Podcasts about OpenCV

O taguaiense Diogo se familiarizou com computadores usando o PC MS-DOS do irmão. Curioso com eletrônica desde pequeno, ele cursou engenharia eletrônica com a intenção de trabalhar com hardware. Ao longo do caminho e, com a experiência de uma iniciação científica nas costas, ele decidiu fazer mestrado em computação gráfica e visão computacional.Foi aí que, graças a um artigo publicado e apresentado na Itália, nasceu a vontade de morar fora. Do fim do mestrado nos Bálcãs, a uma passagem pela França graças ao Ciência Sem Fronteiras, Diogo acumulou experiências que lhe colocaram na rota para trabalhar na área de tecnologias imersivas da Meta.Neste episódio, o Diogo detalha sua curiosa relação de idas e vindas entre a indústria e a academia, e compara suas passagens por outros países da Europa com sua residência atual na terra onde tudo é mais caro.Fabrício Carraro, o seu viajante poliglotaDiogo Luvizon, Engenheiro de Visão Computacional na Meta em Zurique, SuíçaLinks:Institutos Max PlanckMeta Reality LabsConheça a Formação Visão Computacional com OpenCV da Alura e entenda como trabalhar com dados visuais, como imagens e vídeos, através do OpenCV em suporte com outras ferramentas.TechGuide.sh, um mapeamento das principais tecnologias demandadas pelo mercado para diferentes carreiras, com nossas sugestões e opiniões.#7DaysOfCode: Coloque em prática os seus conhecimentos de programação em desafios diários e gratuitos. Acesse https://7daysofcode.io/Ouvintes do podcast Dev Sem Fronteiras têm 10% de desconto em todos os planos da Alura Língua. Basta ir a https://www.aluralingua.com.br/promocao/devsemfronteiras/e começar a aprender inglês e espanhol hoje mesmo! Produção e conteúdo:Alura Língua Cursos online de Idiomas – https://www.aluralingua.com.br/Alura Cursos online de Tecnologia – https://www.alura.com.br/Edição e sonorização: Rede Gigahertz de Podcasts

Discerning and Defining a product manager Role is S.10 E.2 n.142 of the FSG Messaging and Optics Podcast, Wait What Really OK hosted by Messaging and Optics Strategist Loren Weisman. Derrick is the guest on this episode of Wait What Really OK. Together Loren and Derrick dig in to the ins, outs, ups and downs of Product Managers. In this episode, Derrick helps with the discerning and defining when it comes to an effective product manager as well as some red flags to watch out for and many of the attributes to look for. This podcast is raw, real and true. Done in one take, a little EQ and up… Proud of the flubs, the ums and the uhs. This was unscripted and in the moment.  Derrick did not have the questions in advance. Derrick Boudwin is a Qualified Director of Product Engineering with over 15 years experience leading international cross-functional teams, using people-centric strategies to develop software resulting in successful, patented, and disruptive products. Derrick is also versed in the Programming Languages of Python, Bash, Visual Basic, Powershell, SQL, Ruby, Java as well as being familiar with Tools and Technologies that include AWS, GCP, Azure, Tensorflow, Docker, Ansible, Terraform, Jenkins, CircleCI, Git, OpenCV, Pivotal, Jira, and ConfluenceTo talk to Derrick about any or all things Product Manager related or to get some help in your product manager search or assistance in interviewing or reviewing your candidates, email: Derrick@DerrickBoudwin.com *Loren Weisman is a Messaging and Optics Strategist. starting as a session/ghost drummer and then music producer, loren has 700 album credits across major and indie labels as drummer and producer. He then shifted to TV production with credits for ABC, NBC, FOX, CBS, TLC and more including reality shows, infomercials, movies and documentaries. Loren wrote three internationally published and distributed books, including Wiley and Sons, “Music Business for Dummies”, as well as GreenLeaf's “The Artists Guide to Success in the Music Business.” https:/lorenweisman.com/ * © 2024 Loren Weisman / Fish Stewarding Group All Rights Reserved ® ℗ *

This week's show features the latest releases Elixir 1.17.0 and 1.17.1, bringing improved type inference and a new Duration data type. We'll also dive into Phoenix Playground's streamlined approach to single-file applications, José Valim shares another Elixir academic research project taking shape on set-theoretic types for behaviors, and updates from the Igniter project on enhancing code generation and project patching. With a look at how Phoenix and Inertia.js are joining forces and the latest on Nx Scholar's new version for machine learning, this episode is packed with cutting-edge developments in the Elixir community, and more! Show Notes online - http://podcast.thinkingelixir.com/208 (http://podcast.thinkingelixir.com/208) Elixir Community News - https://github.com/elixir-lang/elixir/releases/tag/v1.17.0 (https://github.com/elixir-lang/elixir/releases/tag/v1.17.0?utm_source=thinkingelixir&utm_medium=shownotes) – Release information for Elixir version 1.17.0. - https://github.com/elixir-lang/elixir/releases/tag/v1.17.1 (https://github.com/elixir-lang/elixir/releases/tag/v1.17.1?utm_source=thinkingelixir&utm_medium=shownotes) – Release information for Elixir version 1.17.1. - https://x.com/davydog187/status/1800962252125667748 (https://x.com/davydog187/status/1800962252125667748?utm_source=thinkingelixir&utm_medium=shownotes) – Dave Lucia shared how Elixir 1.17.0 helped find a bug using new type information. - https://x.com/josevalim/status/1801000076497539482 (https://x.com/josevalim/status/1801000076497539482?utm_source=thinkingelixir&utm_medium=shownotes) – José shared additional academic research being started on Elixir for set-theoretic types - https://x.com/wojtekmach/status/1802975489230811354 (https://x.com/wojtekmach/status/1802975489230811354?utm_source=thinkingelixir&utm_medium=shownotes) – Wojtek Mach's announcement of "Phoenix Playground" for creating single-file Phoenix apps. - https://dashbit.co/blog/announcing-phoenix-playground (https://dashbit.co/blog/announcing-phoenix-playground?utm_source=thinkingelixir&utm_medium=shownotes) – Blog post about the announcement of Phoenix Playground. - https://github.com/phoenix-playground/phoenix_playground (https://github.com/phoenix-playground/phoenix_playground?utm_source=thinkingelixir&utm_medium=shownotes) – GitHub repository for Phoenix Playground. - https://github.com/inertiajs/inertia-phoenix (https://github.com/inertiajs/inertia-phoenix?utm_source=thinkingelixir&utm_medium=shownotes) – GitHub repository for Inertia.js Phoenix LiveView adapter. - https://inertiajs.com/ (https://inertiajs.com/?utm_source=thinkingelixir&utm_medium=shownotes) – Official site of Inertia.js, explaining its concept and features. - https://github.com/DockYard/flame_on (https://github.com/DockYard/flame_on?utm_source=thinkingelixir&utm_medium=shownotes) – Update announcement for FlameOn from Dockyard, now with SVG download capability. - https://tylerbarker.com/posts/liveview-is-not-a-zero-js-framework-it-s-a-zero-boring-js-framework (https://tylerbarker.com/posts/liveview-is-not-a-zero-js-framework-it-s-a-zero-boring-js-framework?utm_source=thinkingelixir&utm_medium=shownotes) – Blog post explaining why LiveView is referred to as a "zero-boring-js" framework. - https://github.com/membraneframework-labs/ex_vision/ (https://github.com/membraneframework-labs/ex_vision/?utm_source=thinkingelixir&utm_medium=shownotes) – Repository for ExVision, using ONNX bindings for AI model integration. - https://github.com/cocoa-xu/evision (https://github.com/cocoa-xu/evision?utm_source=thinkingelixir&utm_medium=shownotes) – EVision which uses OpenCV bindings. - https://elixir-nx.github.io/axon/onnxtoaxon.html (https://elixir-nx.github.io/axon/onnx_to_axon.html?utm_source=thinkingelixir&utm_medium=shownotes) – Guide on converting ONNX models to Axon. - https://github.com/ash-project/igniter (https://github.com/ash-project/igniter?utm_source=thinkingelixir&utm_medium=shownotes) – GitHub repository for Igniter, aiming to solve issues with composable mix generators. - https://elixirforum.com/t/igniter-a-code-generation-and-project-patching-framework/64181 (https://elixirforum.com/t/igniter-a-code-generation-and-project-patching-framework/64181?utm_source=thinkingelixir&utm_medium=shownotes) – Elixir forum discussion on Igniter framework's capabilities and applications. - https://x.com/josevalim/status/1803040816404849008 (https://x.com/josevalim/status/1803040816404849008?utm_source=thinkingelixir&utm_medium=shownotes) – José announced the release of Nx Scholar v0.3.1, featuring new embedded notebooks. - https://hexdocs.pm/scholar/manifold_learning.html (https://hexdocs.pm/scholar/manifold_learning.html?utm_source=thinkingelixir&utm_medium=shownotes) – Documentation on manifold learning in Nx Scholar v0.3.1. - https://x.com/yevkurtov/status/1800851584827711607 (https://x.com/yevkurtov/status/1800851584827711607?utm_source=thinkingelixir&utm_medium=shownotes) – Yevhenii Kurtov shared a guide on ES/CQRS with EventStoreDB and Phoenix/LiveView. - https://kurtov.pro/blog/2024/06/an-end-to-end-es/cqrs-example-with-eventstoredb-and-phoenix/liveview/ (https://kurtov.pro/blog/2024/06/an-end-to-end-es/cqrs-example-with-eventstoredb-and-phoenix/liveview/?utm_source=thinkingelixir&utm_medium=shownotes) – Blog detailing an end-to-end example of ES/CQRS implementation. - https://github.com/commanded/commanded (https://github.com/commanded/commanded?utm_source=thinkingelixir&utm_medium=shownotes) – GitHub repository for Commanded, a library for CQRS/ES architectures. - https://learn.eventstore.com/an-end-to-end-example-with-eventstoredb (https://learn.eventstore.com/an-end-to-end-example-with-eventstoredb?utm_source=thinkingelixir&utm_medium=shownotes) – Official guide on implementing end-to-end examples with EventStoreDB. - https://x.com/CodeBEAMio/status/1800918581225431318 (https://x.com/CodeBEAMio/status/1800918581225431318?utm_source=thinkingelixir&utm_medium=shownotes) – CodeBEAM conference announcement, set to take place in Berlin. - https://codebeameurope.com/ (https://codebeameurope.com/?utm_source=thinkingelixir&utm_medium=shownotes) – Official site for CodeBEAM Europe conference details. Do you have some Elixir news to share? Tell us at @ThinkingElixir (https://twitter.com/ThinkingElixir) or email at show@thinkingelixir.com (mailto:show@thinkingelixir.com) Find us online - Message the show - @ThinkingElixir (https://twitter.com/ThinkingElixir) - Message the show on Fediverse - @ThinkingElixir@genserver.social (https://genserver.social/ThinkingElixir) - Email the show - show@thinkingelixir.com (mailto:show@thinkingelixir.com) - Mark Ericksen - @brainlid (https://twitter.com/brainlid) - Mark Ericksen on Fediverse - @brainlid@genserver.social (https://genserver.social/brainlid) - David Bernheisel - @bernheisel (https://twitter.com/bernheisel) - David Bernheisel on Fediverse - @dbern@genserver.social (https://genserver.social/dbern)

The 16:9 PODCAST IS SPONSORED BY SCREENFEED – DIGITAL SIGNAGE CONTENT What if you could use AI to make digital signage screen content relentlessly relevant? That's the premise and promise of what Korbyt calls Machine Learning Broadcast, new capabilities in the Dallas-based software firm's CMS platform. Using computer vision and machine learning, the idea is that if the platform can get a sense of what's making people stop and watch in a defined environment, then content can be optimized based on that interest. The system finds and schedules content to push to screens based on engagement metrics. How it all technically works is a bit over my shiny head, but I had a good chat with Korbyt CTO George Clopp about what's going on and its implications. We also get into what the future looks like for AI in digital signage. Subscribe from wherever you pick up new podcasts. TRANSCRIPT Geroge, thank you for joining me. We've chatted in the past. For those who don't know Korbyt, can you give me a rundown of what the company's all about?  George Clopp: Hi, Dave. It's a pleasure to speak with you again. Yeah, Korbyt is at its root an employee engagement company. So we've got roots in digital signage, but our typical use case is using digital signage at corporate campuses and to communicate to employees, to increase employee engagement as well as to communicate real-time mission-critical stats as well.  Is that pretty much the core vertical that you guys chase, workplace?  George Clopp: It is. We are heavily into the workplace, meeting rooms as well. We do a lot with retail banks, a little bit into the retail space, but it's primarily corporate campuses. For those who don't know the company, it actually goes back a long way to Symon Communications days, right? You guys were doing workplace communications long before the digital signage industry discovered that.  George Clopp: Yeah, exactly right, Dave. It precedes me. I've been here for seven years now. I can't even believe it, but that's how much I enjoy this space and the industry. I enjoy the company so much, but we had Target Vision, Symon Communications, and we've just evolved. I joined at the tail end of 2016 to develop the Korbyt platform, and obviously, we have to meet the needs of the digital signage industry, but we've had a really heavy focus on employee engagement as well. Is it interesting to see all these other companies who have more general offers, find their way into the workplace because they see that as an opportune vertical?  George Clopp: Yeah, I view it as exciting. I think it's definitely a macroeconomic trend with the pandemic, post-pandemic, the modern workplace, everything is reimagining and reinventing and re-everything these days. I think it's good. It's a legitimate macro problem that everyone's looking to provide solutions to. So, I'm really excited. I love the industry myself.  In some respects, you guys have been doing back-of-house, a lot longer than most companies would have. I mean, you're not just working in the offices, you're working in production areas and so on.  George Clopp: That's correct. Heavy in manufacturing and heavy in the contact centers, anytime where you're doing mission-critical real-time data, you're connecting to an ERP (Enterprise Resource Planning), or yard management system, and you want to change or orchestrate the display and the surroundings based on data changing, we've got a deep background in that.  Yeah, for contact centers, if I recall, years ago pre-arrival with the company, you were doing low-resolution LED readouts that were just telling people in the contact center about the average wait time on calls and things like that.  George Clopp: Exactly, and that's matured over the years and now we're doing that on the desktop and on the mobile device as well. We still have some supply chains and some yard management systems in a warehouse, where we'll do the little blinky boards over the dock doors themselves. We range from the dock doors all the way to your mobile device now.  The PR that came out about a new piece of functionality, your marketing talks about a million endpoints, 250 cloud migrations, and 100+ native integrations.  A million endpoints, that's like a lot. George Clopp: It is. Yeah, scalability and being able to expand out to touch desktops, normal, typical digital science screens, and mobile endpoints. It's been a real focus on us for the last four or five years. So we're really proud to announce that, and then the back end, like you were talking about those native data integrations, I think that's really what sets us aside from a lot of our competition is making those really hardcore authentications and then that real-time pipe between us and the source systems.  I know a lot of other software in our space that we run into, they talk about integrations. A lot of times it's really just a file, they're taking data from a source system. They're putting it into a CSV format or any kind of other format and then they're pulling that in. So that's really where we shine with that real-time data integration.  Is that important in terms of a distinction when solutions providers and users are looking at data integration and they see that a CMS says, yeah we do data integration, we can integrate with your platform? It sounds like you're saying there are different tiers of that, and there's real integration and there's just like a baseline.  George Clopp: Yeah, exactly. That's the right way to pick up on that day, for sure. When you need to orchestrate and change things in a 911 center or in a manufacturing-type environment and definitely in a contact center, speed is really the key there. So having something on a five-minute loop that's pulling a file, it's just not fast enough. So you need that real-time data, you need that high availability so that something was to break that you've got a backup in place and you can make sure that contact center, that supply chain, that 911 center is rolling smoothly. They're not just getting their data, but they're changing the experience of the data. That's another thing that we do, we pull in stats, but we also augment those stats and do value-added calculations on the stats, and then we trigger on those values to change the screen, or change the mobile device or change the desktop. So if you've got too many calls in the queue or you're running behind on this loading dock here, we'll change the entire experience for you based on that value-added stat that we do.  I also assume that when companies talk about integrations, for very logical reasons, they're going to go to the most used platforms out there, whether it's Teams or God knows what. But if you have a hundred plus native integrations you're probably talking about some pretty exotic things that nobody's ever heard of, and if a company went in and said, we can integrate with their systems and they say, what those systems are, their eyebrows are going up, because they're thinking, I have never heard of that. George Clopp: Absolutely, Dave. There are some low-level protocols where we just integrate at a TCP level with a very proprietary protocol, but I would say the bulk of it is more modern, JSON-based RESTful interfaces, for sure and we like to distinguish between data integrations, business application integrations, and SSO integrations, in three categories there. So, like a Power BI or a Tableau or something like that would be more of a business application integration, and when we're talking data integration, we're talking more low level, running SQL against a data store, running web services, running SOAP-based web services, and to that extent. And again, that's why we call it out in our marketing because we do think that's a core differentiator for us.  So just to go back to something, when you talk about a million endpoints, you're including desktops..  George Clopp: That's correct. Desktops and mobile devices, basically all of the endpoints that we talk to. Good. Back at the start of summer, you guys introduced something called, Machine Learning Broadcast. What is that? George Clopp: Yeah, fantastic question. We were involved with machine learning, and AI before it was really cool, so this was actually something we developed in 2018. We've been honing the model, and then we re-released it this year. But machine learning is a subset of AI, and we all know AI is a super big buzzword these days and when you peel that onion, there's levels of accuracy involved there, and there's a lot of hype around the world. But the reason why we called the feature machine learning broadcast is really to focus on the ML aspects of it, and it's a great business problem to solve because, at the end of the day, what we're really creating is a recommendation engine. And I think everybody's familiar with the Amazon recommendation engine, Instagram, and other social media platforms that are just, they're recommending content for you. That's essentially what we're doing here. We're using KNN Analysis, which is supervised machine learning to look at content that has some engagement with it, and that engagement could be measured by computer vision on a digital signage screen, it could be measured by interactivity with it on a desktop or interactivity with that content on the mobile device and then behind the scenes, all we're doing is we're finding out second, third, fourth-degree order content, that's related to the content that was engaging and then it's a feedback loop. We go ahead and automatically schedule that content and see how that content is engaged with so it's a self-learning feedback loop there and the whole purpose of it is to find content that's engaging and show more of that content to your employees. Could you give me a real-world kind of example of how that might work? George Clopp: Yeah, absolutely, Dave. Let's say a company's opening up a brand new office in Buenos Aires and for whatever reason, people really gravitate to that content. They look at it on the signage screen, on the fifth-floor break room, they're engaging with it on their desktop, they're looking at it on the mobile device. We learn from that engagement and say, okay, let's go ahead and find similar related content there. Let's find content related to office openings in Buenos Aires, and then let's go ahead and go further out and look at second, third-order tags. So that would be content related to South America as well. And then we automatically play that content, inject it back into the playlist, and our customers have complete control over whether it's automatic and which players actually get this content and which devices get it and then, we learn based on that content. So it's a feedback loop, and you might find in that case that your employees are really more interested in the geographic region than they are in the new office opening. So it's relentlessly relevant.  George Clopp: Exactly right, Dave, and solving a real-world business problem because one of the challenges our customers have is, it's really arduous to constantly schedule new relevant content.  The first couple of times you do it, you create a scheduled playlist. Yeah, it's okay, but it takes a long time and then, with Attention Deficit Disorder in today's modern world, people grow immune, and they tune out that same content over and over again. So, you need that fresh content injected to keep the employee's attention.  I'm guessing that somebody's going to be listening to this and thinking, that's cool, but where on earth do I get, or how do I develop all this content so that I do have this somewhat bottomless hyper-relevant content available? George Clopp: Yeah, fantastic question. Right now, in its current stance with our ML broadcast, you need to have that content in your media library. We're not automatically going out to like copyright-free areas and pulling in content. But with our release coming out next year, it's called our AI employee engagement. With that, we'll automatically be creating and sourcing content for you on your behalf.  Yeah, I saw a demo of something like that over in Germany a little while back with another company who, I'm sure you'll be happy if I don't name them, that was all about using what was available through an intranet and an extranet, and other resources to auto-generate content for screens. George Clopp: Yeah, it's opening up the whole world of generative AI. We're actually looking at both. Whether there are generative images, generative video, or generative text. Obviously, in our space, images and videos mean a lot, and there are different systems out there. There's DALI 2, there's stable diffusion. They've all got their strengths and their weaknesses. But we're combining that with templated-based content as well.  So automatically generating content that's relevant based off of a text prompt is super useful. But in some cases, it might not be the right content that's generated. So we also will have a mixture of templated content as well.  Yeah, I think templates are a big part of that. I've farted around with things like Mid Journey and so on, and you could see how it could go sideways on you really quickly if you left too much up to the machine.  George Clopp: Exactly. It gets into that whole thing of prompt engineering. You got to be really good with your prompts, and they've all got issues like generating hands and things of that nature right now. But we want to be on the leading edge of this, use it where it makes sense. An area where we think it really makes a lot of sense, a preview into our AI Employee Engagement, is on mission values and goals. We feel like that's an area where our customers just don't communicate enough to their employees, like, there's cake in the break room, let's recognize employees.  That's all part of it, but really just reinforcing, Hey, your goal in the finance department this week is to close your books three days earlier. And so, mix that text in with some great video or some great images that are created in the background using this generative AI. Yeah, I saw something on LinkedIn last night, and I commented on it because I thought it is great that there's a company that's using KPIs and messaging right on the production floor, and the person who posted about it said, this is not very sexy, but it goes to what's needed on the floor for those workers. But the problem was, it looked like hell.  It was just black and white, and they were slapping up a whole bunch of Excel charts, like a stock of them and you'd need binoculars to even see them. So it's important to think about the presentation.  George Clopp: Yeah, totally agree, Dave. I say this at all my speaking events: content is king, content is queen, and that still rules the day. When we're intermixing real-time data with content, it has to be visually appealing. You can't have 20 different stats on the screen; all of those rules of graphic design, I still think, hold true here.  Do you see a day when things like scheduling and trafficking of content are largely automated and handed off to machine learning or some variant of AI?  George Clopp: That's exactly what we're trying to build, Dave, with a release next year. With the ability, of course, to intervene, the ability for the communicator to come in and approve the content or really go ahead and bias the content and say, okay, I've got these 30 categories of content I see that I really want to bias, what the content areas could be.  “Hey, I'm a new enroll. I'm a new first-time line manager. I'm a new director. I'm a new VP, and there's content associated with that new enroll.” They might want to bias that and increase the weight on it, decrease the weight on it, or take it out altogether. So there's still going to be that human touch involved in the ability to approve content, but the AI itself will take care of making sure that content is fresh and relevant. And the big problem we're solving there is just that, again, attention deficit disorder people have, if they see the same thing on the screen, week after week, they tend to tune out. So how can we think of innovative ways to display KPIs, display goals, display things that are really important to the company and give it a great background, give it a great video so that it gets employees' attention again? We're going to talk about machine learning. You reference AI-driven camera optics. Is that basically a computer vision? George Clopp:  It is. Absolutely is, yes. Did you guys write your own, or are you using something like Intel's OpenVINO?  George Clopp: Yeah, the two big ones out there, we've used OpenCV, that is, Open Computer Vision, and TensorFlow, and they both have their strengths and weaknesses, but there are higher order problems we're trying to solve here, and not reinvent computer vision so we're using some libraries for that.  Is that just part of the mix of doing this sort of thing? Are there other technologies you can use to get a sense of dynamics in a venue? George Clopp: Yeah, I think so. Infrared detectors, pressure sensors that kind of tell you who's in that immediate vicinity. You're basically correlating that to human beings in the vicinity, how many human beings are there, and what was playing on the screen at that time. Yeah, so there are less technological ways to do this and still get some good results.  AI is being talked about a lot as you've gone through about its potential to automate presentations. Are there other aspects to a digital signage company, the way your company operates, that you can use AI to help with marketing, help with customer contact, that sort of thing? George Clopp: Yeah, without a doubt. I'm sure you're reading everything. It's revolutionizing all traditional roles, right? Not just engineers writing code. You got a chat with a ChatGPT engineer. With Microsoft's Copilot, it's going to revolutionize the way we all use Excel and Word and PowerPoint and things of that nature. It's definitely revolutionizing marketing. Building product brochures for you automatically, things of that nature, and then, that naturally progresses into, is AI going to take all of our jobs, which I don't think so, going to help us all become more productive. The employees that really change and adopt the AI, I think they're going to be even more valuable than they are today. It's just the employees that just say, I'm not going to do this, and they refuse to allow their cheese to be moved, those are the ones that I think you have to watch out for.  There's an increasing number of companies. I just wrote about one today that has gone down the path of headless CMS. The idea that you can leave the final presentation later, the interactive element, whatever it is to software developers at a large company or who works with a large company as a services company and the digital signage CMS is just the infrastructure, the foundational platform that does device management, scheduling, trafficking, all that sort of stuff. Are you seeing that demand in the marketplace?  George Clopp: We're seeing the opposite. What you're saying absolutely makes sense, especially with my background and the way we've architected our product with microservices. What we're seeing, especially with our large enterprise customers is, they want a little more white glove service. Taking on the arduous task of piecing everything together, even with a microservices framework, is putting a lot of ownership on them. But that is not to say that there's not a need out there. We just really haven't found it. We've actually gone the opposite direction on our side, which has really served us well because we've gone from zero revenue in the cloud to 2 million. We brought on a new CEO, and we quickly ramped up to 20 million. I think it's working for us so far.  Yeah, you're a very different company than maybe prior to you joining RMG Networks, that was a weird little side trip into digital out of home.  George Clopp: It was. We see the artifacts and all that, but I think it's a great group of people here now. There's not a leftover where people have bad attitudes or anything like that. So really proud of where the company's been, the talent we've acquired. We've acquired people from all over the industry. Really love working with the current team and cross-functionally, not just engineering and support, which is what I run, but in sales and marketing as well.  Yeah, it's interesting when you mentioned you've gone in the opposite direction of headless. I've heard that as well, particularly when you get into, like Fortune 500, Fortune 100 kinds of enterprise-grade customers. They want to outsource digital signage, by and large, in the same way that they've outsourced a lot of IT services. George Clopp: Yeah, absolutely. That's the same trend we're seeing, Dave too. It's a little bit of both, right? Everybody wants their cake and eats it too, right? Like they want you to have the ability to do it, but then when it comes time to actually execute on it, we typically find, Hey, we can help them get faster to market if we help augment their team. How important is security? George Clopp: Oh! It's Huge. We all know that the disaster scenario in digital signage, someone compromises your network and they put up some content images or videos that are not appropriate. Even more so with us being more omni-channel with desktop, mobile devices. We've got a data privacy officer, we're SOC 2 compliant. We do a lot of work in Europe so GDPR comes up a lot as well, data privacy. So I think it's super important.  When I think you look at the different offerings out there and the first tier, we look and sound the same. So I think what you got to do with new prospects or new customers, they just got to peel that onion more. What does that really mean? What does it mean that you encrypt your data? Do you do it at rest? Do you do it in transit? Those kinds of things, and I think that's where you can tell the difference between different offerings.  And are the people in the first and second meetings with prospective customers different than they were 7 years ago when you started? I'm hearing the IT people who used to come to meetings and sit there with their arms crossed, thinking, dear God, how long is this going to go on? They're now tending to lead these meetings.  George Clopp: Yeah, I've seen it in multiple ways. Definitely, IT is still the big persona of the buyer here. But I'm also seeing less and less about speeds and feeds and players and hardware and transmission equipment and scalers and more about the final purpose of what we're trying to do. I'm just starting to see that shift. Seven years ago, I talked to people, and it's the AV integration guy. I don't really care what's on the screen. I just care that it's not dark. I don't want a screen that's down. That's their most important thing, and now I'm seeing that shift a little bit more towards they do care about the content, and they're bringing in more of the HR and the communications group involved and making sure that the platform can grow. I can create content on the platform or I can integrate with Adobe or SharePoint or something along those lines. But I still see it, especially AV/IT as a huge influence in the buying process.  Yeah, certainly going back seven, eight years when I was doing some one-to-one consulting with enterprise level customers, that sort of thing, I would go into a first meeting, and I would say, okay, why do you want to do this? And it was always intriguing to see how often people would lean back in their chairs and say, I hadn't really thought about that. They wanted this thing, but as you say, they didn't really know what they were going to do with this thing. George Clopp: Yeah, exactly. And there's a little bit of power in that too. There's power to putting the latest and greatest screen technology in your office and giving you that modern technology look and feel but then just carry it one more step in the maturity direction and start focusing on the content too. Yeah, you can demonstrate innovation by having a big ass screen in your lobby, but if there's nothing useful on there, you're not really demonstrating a lot of innovation.  George Clopp: Exactly, and I think there's still room for that super wonderful creative experience that's human-curated that graphic designers make, and they spend a lot of time getting just perfect in those high profile areas, like the lobby of a company, and then there's also opportunity for, new content generation automatically for me so that I don't have to necessarily sit here and handle this thing. So I think we're going to live in a world where both will be applicable. So you mentioned you, you're working on new iterations of AI-driven content. Is that the big kind of roadmap item for your company over the next year?  George Clopp: Yes, it really is. Yeah. We've got a huge, large-player ecosystem, all the data integrations, and omni-channel platforms. So where our new development team is focused on is automating the content creation, automating that entire feed, if you will, so that it really takes that arduous process away from our communicator. How many folks do you have in the company now? George Clopp: We're a little under 70 people right now. So still a small company and I love it cause everybody has to wear multiple hats, do multiple roles. You have to bring a lot of energy to the company, and I just love that. I've just grown so fond of it over the last seven years.  And is most of the team in the Dallas Fort Worth area, or are you all over the place? George Clopp: Since COVID, we're mainly in Dallas, but since COVID, a lot of us have moved out a little bit. So I'm actually in Colorado. Some of my engineering leads are in the West Coast, some are in Pennsylvania. So we're really practicing what we preach, the hybrid workforce. All right, George, thank you for spending some time with me. It was good to catch up. George Clopp: Yeah, it's fantastic, Dave. Thank you so much for taking time out.

Ini proses saya ngoprek OpenCV di MacOS. Awalnya gagal. Dilanjutkan videonya setelah berhasil.

We debate if users learned their lesson from the Docker Hub drama, and the silent self-hosting winner going from strength to strength. Proxmox gets some big updates. Plus, our thoughts on the state of self-hostable AI tools.

En este episodio se habla sobre diversas herramientas de generación de imágenes con inteligencia artificial, que permiten un mayor control en la generación de imágenes y posiciones de personajes. Se mencionan las siguientes herramientas: Control Net: tecnología de preprocesamiento que analiza una imagen de referencia para generar imágenes con mayor control. Utilizada principalmente para métodos imagen a imagen. Open Pose: tecnología de OpenCV que analiza la pose y postura de una persona en tiempo real, generando un esqueleto en líneas que permite posicionar al personaje. Magic Poser: aplicación para móvil y escritorio que permite arrastrar nodos del cuerpo para posicionar al personaje. Utiliza cinemática inversa. Pose Maniacs: herramienta web que permite seleccionar entre avatares masculinos y femeninos para posicionarlos en distintas poses predefinidas. PostMD: interfaz web que permite crear escenas con múltiples personajes, seleccionar poses y añadir modelos y pros. Versión gratuita y versión pro con más opciones. Daz Studio 3D: aplicación gratuita de escritorio que trabaja con modelos 3D y permite controlar la iluminación y vestuario. Tiene una curva de aprendizaje más complicada, pero ofrece mucho más control. No te olvides de seguir el canal de youtube en redes sociales y a utilizar el servicio de asesoría tecnológica personal. OpenPoser Editor -> Para Automatic 1111 -> https://arxiv.org/pdf/1812.08008.pdf Magic poser -< https://magicposer.com/ posemaniacs -> https://www.posemaniacs.com/ posemy.art -> https://app.posemy.art/ DazCentral -> https://www.daz3d.com/ Plan de Asesoria Personal Telegram Tecnolitas Déjame un mensaje de voz

En el episodio de esta semana seguimos con ideas de proyectos sobre inteligencia artificial que podrás realizar por tu cuenta con estas pequeñas guías de ideas, atrévete a ser un ingenios@ de sistemas y poner en practica tus conocimientos 6. Modelo de reconocimiento de gestos de la mano Puedes crear una aplicación web de reconocimiento de gestos en Python. Para ello, puedes utilizar la base de datos de reconocimiento de gestos de la mano en Kaggle. Este conjunto de datos consta de 20.000 gestos etiquetados. Puedes entrenar este conjunto de datos en VGG-16. También puedes utilizar OpenCV para recoger un flujo de datos de vídeo en directo y utilizar el modelo para detectar y hacer predicciones sobre los gestos de la mano en tiempo real. Incluso puedes crear una aplicación de reconocimiento de gestos de la mano. Despliega tu modelo en un servidor y deja que haga predicciones a medida que los usuarios hacen una variedad de gestos con las manos. Dataset: Kaggle Hand Gesture Recognition 7. Modelo de generación de texto GitHub: GPT-3 8. Detección del color Dataset: Kaggle Color Recognition Dataset 9. Aplicación de reconocimiento de la lengua de signos con Python Dataset: World-Level American Sign Language dataset 10. Detección de la violencia en los vídeos Datasets: Violent Flows Dataset / Hockey Fight Videos Dataset Déjame un mensaje de voz

La lutte contre les deepfakes compte aujourd'hui un nouvel acteur de taille : Intel ! Mi-novembre, le géant américain des semi-conducteurs a présenté FakeCatcher, son intelligence artificielle capable de détecter en temps réel des vidéos truquées. C'est dans un communiqué de presse qu'Intel a dévoilé FakeCatcher, une plateforme dont l'objectif est de devenir, je cite le « premier détecteur de deepfakes en temps réel au monde qui renvoie des résultats en quelques millisecondes » fin de citation. Dans le détail, FakeCatcher a été conçu par Ilke Demir, chercheur chez Intel Labs, et Umur Ciftci, de l'Université de l'État de New York. L'ossature de FakeCatcher est composée de plusieurs outils et logiciels développés par Intel, comme OpenVino, Intel Integrated Performance Primitives et OpenCV. Ainsi, ce nouveau dispositif d'Intel basé sur la technique du deeplearning, apprentissage profond, se distingue des autres technologies de détection des visages par sa capacité à analyser, je cite « des blocs de vision ». En s'appuyant sur l'Open Visual Cloud et les processeurs Intel Xeon Scalable 3e génération, les développeurs ont gagné en vitesse, c'est-à-dire que FakeCatcher peut traiter simultanément jusqu'à 72 flux détectés dans un pixel vidéo. Voilà grossièrement résumé le côté technique de ce nouvel outil. Ceci dit, si vous n'êtes pas familier avec les deepfakes, sachez que ces vidéos sont généralement montées par des personnes mal intentionnées, qui font notamment tenir des propos outranciers à leurs victimes. L'ancien président des États-Unis Barack Obama a notamment servi d'exemple il y a quelques années, tout comme l'opposant à Vladimir Putin, Alexeï Navalny, dont le visage avait été détourné pour servir la communication du gouvernement russe il y a quelques années. D'après le cabinet Gartner, les dépenses des entreprises liées à la cybersécurité s'élèveront à pratiquement 190 milliards de dollars en 2023, soit une hausse de plus de 11 % par rapport à cette année. À noter qu'Intel n'est pas le seul sur le marché des outils de détection de deepfakes. Facebook et Alphabet s'y essayent déjà depuis plusieurs années, quand Microsoft a fait son grand saut en septembre avec le Video Authenticator. Ceci dit, il convient de ne pas diaboliser totalement les deepfakes, étant donné que cela peut aussi servir, notamment au cinéma ou dans les séries pour faire apparaître un acteur décédé ou en rajeunir un autre par exemple. Learn more about your ad choices. Visit megaphone.fm/adchoices

La lutte contre les deepfakes compte aujourd'hui un nouvel acteur de taille : Intel ! Mi-novembre, le géant américain des semi-conducteurs a présenté FakeCatcher, son intelligence artificielle capable de détecter en temps réel des vidéos truquées.C'est dans un communiqué de presse qu'Intel a dévoilé FakeCatcher, une plateforme dont l'objectif est de devenir, je cite le « premier détecteur de deepfakes en temps réel au monde qui renvoie des résultats en quelques millisecondes » fin de citation. Dans le détail, FakeCatcher a été conçu par Ilke Demir, chercheur chez Intel Labs, et Umur Ciftci, de l'Université de l'État de New York. L'ossature de FakeCatcher est composée de plusieurs outils et logiciels développés par Intel, comme OpenVino, Intel Integrated Performance Primitives et OpenCV. Ainsi, ce nouveau dispositif d'Intel basé sur la technique du deeplearning, apprentissage profond, se distingue des autres technologies de détection des visages par sa capacité à analyser, je cite « des blocs de vision ». En s'appuyant sur l'Open Visual Cloud et les processeurs Intel Xeon Scalable 3e génération, les développeurs ont gagné en vitesse, c'est-à-dire que FakeCatcher peut traiter simultanément jusqu'à 72 flux détectés dans un pixel vidéo. Voilà grossièrement résumé le côté technique de ce nouvel outil.Ceci dit, si vous n'êtes pas familier avec les deepfakes, sachez que ces vidéos sont généralement montées par des personnes mal intentionnées, qui font notamment tenir des propos outranciers à leurs victimes. L'ancien président des États-Unis Barack Obama a notamment servi d'exemple il y a quelques années, tout comme l'opposant à Vladimir Putin, Alexeï Navalny, dont le visage avait été détourné pour servir la communication du gouvernement russe il y a quelques années. D'après le cabinet Gartner, les dépenses des entreprises liées à la cybersécurité s'élèveront à pratiquement 190 milliards de dollars en 2023, soit une hausse de plus de 11 % par rapport à cette année. À noter qu'Intel n'est pas le seul sur le marché des outils de détection de deepfakes. Facebook et Alphabet s'y essayent déjà depuis plusieurs années, quand Microsoft a fait son grand saut en septembre avec le Video Authenticator. Ceci dit, il convient de ne pas diaboliser totalement les deepfakes, étant donné que cela peut aussi servir, notamment au cinéma ou dans les séries pour faire apparaître un acteur décédé ou en rajeunir un autre par exemple. Hébergé par Acast. Visitez acast.com/privacy pour plus d'informations.

“But here's my plug: All the great things about Christmas are present at Thanksgiving - without the pressure of buying shit for people. And so you get the feast. You get the opulent … the harvest feast. And you're usually connecting with an N of people that you have an N amount of feelings for. And maybe some people are struggling with their family - I get that.You're usually connecting with an extended group of people you care about and there's no pretense… It's just like, “Yeah, we gotta have this meal together.” Worst case scenario, you get into a turkey measuring contest. It happens. Not everyone can wield a 22-pound turkey.”Show Notes:Trust fails (00:46)VRBO sucks (04:51)The ship of Theseus (09:13)AirBNB stories (10:23)New York taxis (13:14)City design (17:42)The dad transition (22:26)It's the holiday season (24:14)The Christmas plug (28:05)Big Chris facts (31:23)Paul Rudd (39:48)Smoking and grilling misadventures (44:13)A mustard conundrum (47:24)Awkward dreams (49:12)Casey discovers PDAL (53:19)Photogrammetry (54:36)OpenCV (56:20)YOLO v3 (57:34)Thanksgiving reminders (1:03:45)Show Links:Lofty wants you! Check out our open positions!PDAL: https://pdal.io/en/stable/Photogrammetry: https://en.wikipedia.org/wiki/PhotogrammetrySupport Friday Afternoon Deploy Online:Facebook | Twitter | Patreon | Teespring

In this episode of Info Product Mastery, we’ll discuss whether or not building a community helps sell info products and online education courses Topics covered [00:32] This week's featured listener review. [01:01] Today’s topic comes from a listener, Misha Manulis. “Just found your podcast via Startups For The Rest Of Us podcast. Loved your PyImageSearch site when I was learning OpenCV years ago.I'm curious about your thoughts / experience with building a community around info products.My goal with looking at this is to understand how to build a community as part of a series of info products. I'm in the middle of building an online school for IoT. I'm frustrated with the industry and its failures over the last 10 years. I want to share my knowledge and experience building IoT products for hobby, B2B and B2C markets. There is so much snake oil and marketing, I want people to have the tools to build their own or to understand what they're buyingMy intuition says that a community around this content would be incredibly valuable in accomplishing these goals.Would love to "hear" any and all thoughts / feedback you can share.” [02:20] “If content is kind, then community is king.” - Adrian [03:02] If you are creating high quality content, then your listeners or readers are going to learn from you. But what happens when they have completed all their goals? [03:22] It’s important to build a relationship with your audience. To build “super customers.” [03:44] The personal connection with your customers is very important. Adrian explains how to do that. [04:28] “People don’t buy brands, they buy the transformation.” - Adrian [05:01] Community is also helpful with supporting your info products. As you grow, you will get customers asking questions about what they are learning and how they can apply it. [05:57] Adrian believes it’s your responsibility as an educator and a mentor to help as much as you reasonably can without doing the work for your customers. [06:05] If you want to build a community, you have to put some time in. However, as the business grows, you will receive more support questions, which can be a burden. Adrian discusses how to deal with this. [06:54] Adrian talks about super customers and how they can help you build community and why that is a good sign. [07:49] It’s key to participate in the community around your info products and support, and also to

Data Scientist at World Wide Technology, Leah Ellis-Clemons gives an overview of the specific considerations for implementing machine learning models in Python in real time including OpenCV and Multiprocessing... She gives us a live demonstration of her code as well! Resources: https://github.com/LEllisClemons/Lego_Detector_Realtime https://www.kaggle.com/code/databeru/classify-bricks-compare-transfer-learning-model/notebook https://www.kaggle.com/datasets/joosthazelzet/lego-brick-images https://www.linkedin.com/in/leah-ellis-clemons-6230aa68/ https://www.udemy.com/course/python-for-computer-vision-with-opencv-and-deep-learning/

Remember that we mentioned computer vision in the last episode of the Digital Orthopedics Podcast? We are talking a little more in depth about it and its future with Gary Bradski, the Chief Scientific Officer at OpenCV.ai.  OpenCV is one of the world's most visited open libraries, with 200,000 downloads per day on various topics. Even though OpenCV is used in robotics and the medical field, they are also used for sports, body motion, agriculture, and many more fields of work. Gary explains how they are looking forward to doing predictive, counterfactual, and explanatory models and more accurate screenings in less time with fewer views.  Gary ends this with his thoughts on causal simulation concerning general AI and how humans perceive the world.  Join the revolution and listen to this episode with one of the best in the industry! 

Watch the live stream: Watch on YouTube About the show Sponsored by us! Support our work through: Our courses at Talk Python Training Test & Code Podcast Patreon Supporters Special guest: Nick Muoh Brian #1: picologging From a tweet by Anthony Shaw From README.md “early-alpha” stage project with some incomplete features. (cool to be so up front about that) “Picologging is a high-performance logging library for Python. picologging is 4-10x faster than the logging module in the standard library.” “Picologging is designed to be used as a drop-in replacement for applications which already use logging, and supports the same API as the logging module.” Now you've definitely got my attention. For many common use cases, it's just way faster. Sounds great, why not use it? A few limitations listed: process and thread name not captured. Some logging globals not observed: logging.logThreads, logging.logMultiprocessing, logging.logProcesses Logger will always default to the Sys.stderr and not observe (emittedNoHandlerWarning). Michael #2: CheekyKeys via Prayson Daniel What if you could silently talk to your computer? CheekyKeys uses OpenCV and MediaPipe's Face Mesh to perform real-time detection of facial landmarks from video input. The primary input is to "type" letters, digits, and symbols via Morse code by opening and closing your mouth quickly for . and slightly longer for -. Most of the rest of the keyboard and other helpful actions are included as modifier gestures, such as: shift: close right eye command: close left eye arrow up/down: raise left/right eyebrow … Watch the video where he does a coding interview for a big tech company using no keyboard. Nick #3: Is Google's LaMDA Model Sentient? authored by Richard Luscombe (The Guardian) The Google engineer who thinks the company's AI has come to life Transcript of conversation Brian #4: richbench Also from Anthony “A little Python benchmarking tool.” Give it a list of (first_func, second_func, “label”), and it times them and prints out a comparison. Simple and awesome. def sort_seven(): """Sort a list of seven items""" for _ in range(10_000): sorted([3,2,4,5,1,5,3]) def sort_three(): """Sort a list of three items""" for _ in range(10_000): sorted([3,2,4]) __benchmarks__ = [ (sort_seven, sort_three, "Sorting 3 items instead of 7") ] Michael #5: typeguard A run-time type checker for Python Three principal ways to do type checking are provided, each with its pros and cons: Manually with function calls @typechecked decorator import hook (typeguard.importhook.install_import_hook()) Example: @typechecked def some_function(a: int, b: float, c: str, *args: str) -> bool: ... return retval Nick #6: CustomTkinter A modern and customizable python UI-library based on Tkinter. Extras Michael: OpenSSF Funds Python and Eclipse Foundations - OpenSSF's Alpha-Omega Project has committed $400K to the Python Software Foundation (PSF), in order to create a new role which will provide security expertise for Python, the Python Package Index (PyPI), and the rest of the Python ecosystem, as well as funding a security audit. (via Python Weekly) Nick: Terms of Service Didn't Read - Terms of Service; Didn't Read” (short: ToS;DR) is a young project started in June 2012 to help fix the “biggest lie on the web”: almost no one really reads the terms of service we agree to all the time. Joke: Serverless A DevOps approach to COVID-19

OpenCV is an open-source platform that's equipped with hundreds of powerful features. From simple image processing to complex machine learning, OpenCV is optimized for performance.  It's possible to get away with as little as 10 MB of RAM for a real-time application. OpenCV is backed by a very supportive community that's always standing by to answer questions, offer solutions, and offer help.  At Dev.co., it's our mission to help clients develop OpenCV solutions that meet their needs in an efficient and cost-effective manner. More info about OpenCV development services:   https://dev.co/opencv/   Connect with us:  SEO // PPC // DEV // WEBSITE DESIGN

In this episode, Adrian discusses the fundamentals of market research and how you can “niche down” to find an audience for your online education business. Topics covered [0:48] Market research boils down to three things: Are people searching for it? Will people pay for a book or course on it? And how competitive is the space? [00:58] Developers love writing code, but sometimes become too focused on the product and not the marketing. [01:59] We are used to having a sales team with us to help guide the product so that we are building a product that solves a problem for a customer. [02:20] One of the worst things you can do is to develop a product without enough market research. [02:52] What you want to be is a solution to an existing problem rather than a solution in search of a problem. [03:13] You are looking for the sweet spot between interest in a topic but not so much interest that it's too tough to break into. Adrian explains what happens in a market that is too crowded. [04:37] Barrier to entry is a key phrase when searching for a topic, and building a business. [05:22] Adrian gives a personal example of his experience with PyImageSearch, and how a barrier to entry gave him an advantage in his niche. [06:46] PyImageSearch was able to be successful with very few competitors for a number of years because of the barrier to entry with learning the OpenCV library. [06:59] Adrian explains how to build a barrier to entry and gives an example of a situation a developer might be able to leverage. [08:53] You want to overlap your experience, knowledge, and ability to create a Venn diagram. At the center of this will be your potential niche. [09:46] Adrian leaves us with some homework. Re-listen to episode four. Then brainstorm new ideas for your info product business while keeping in mind the Venn diagram concept. Links from the show PyImageSearch OpenCV library More information on “barriers to entry” If you have any questions from this episode, or have a question you want me to answer on the show, please submit your question via our contact form. We'd love to hear from you.

Join Hackaday Editor-in-Chief Elliot Williams and Managing Editor Tom Nardi for a recap of all the best tips, hacks, and stories of the past week. We start things off with an update on Hackaday's current slate of contests, followed by an exploration of the cutting edge in 3D printing and printables. Next up we'll look at two achievements in detection, as commercial off-the-shelf hardware is pushed into service by unusually dedicated hackers to identify both dog poop and deep space pulsars (but not at the same time). We'll also talk about fancy Samsung cables, homebrew soundcards, the surprising vulnerability of GPS, and the development of ratholes in your cat food. Check out all the sweet, sweet links over on Hackaday.

AI is a fascinating technological pursit, but also a confusing one. Mostly used as a marketing term the AI landscape has become confusing especially for beginners. In this interview we ask Dr. Satya Mallick, CEO of OpenCV all your AI questions. Did you know, for example, taking panorama photos or using HDR is computer vision, but not AI? Check out the show notes at: https://podcast.soar.com/uplift-humanity-podcast/satya-mallick/ Get 10 free hours of AI Video Search tech on your website at soar.com/deepsearch

AI has turned into a marketing term, and the result is confusion for everyone on what AI is. So, what is AI, and what isn't AI? For example, did you know taking panorama photos or using HDR is computer vision, but not AI?  The difference is if the system, or machine, is learning through data, or if the process is fixed. Face recognition requires AI for example.  What is OpenCV for? Why do we need an open-source library for AI? In a way they are collecting the lego blocks, so builders can focus on building new things, not starting from scratch. Go to JustinBradyShow.com for the full show notes and clickable highlight chapters. 

We talk with Cocoa Xu about his PhD robotics project and his related Evision project that creates Elixir bindings to the OpenCV library. The project enables computer vision in Elixir much easier by building on existing projects. We learn about the kinds of features this enables and how it can target embedded devices as well. His goal of a clustered, collaborative, hoard of Elixir robots is terrifyingly fascinating!

С наступающим Новым Годом. Обычно в конце декабря мы проводим ретроспективу уходящего года, но в этот раз мы не будем делать "Голубой Огонек". Вместо этого мы пригласили интересного гостя, которая расскажет как работают алгоритмы распознавания изображений и ML. Мы часто экспериментируем и нам очень важно Ваше мнение. Поделитесь им с нами в опросе: https://forms.gle/vAb2rN6MhTK71YMN9 Спасибо всем кто нас слушает. Не стесняйтесь оставлять обратную связь и предлагать свои темы. Shownotes:  0:03:45 Про нюдс фильтры 0:16:05 Свертка 0:22:30 Сверточные нейронные сети 0:31:30 Распознавание без ML 0:44:30 AR и QR коды 0:57:40 Готовимся к экзамену по цифровым изображениям за 20 минут 1:22:50 Про JPEG 1:40:00 Библиотечки 1:49:00 Новости индустрии обработки изображений 2:00:00 С Новым Годом Ссылки: - https://github.com/SixLabors/ImageSharp : ImageSharp  - https://imagemagick.org/ : ImageMagick - https://github.com/dlemstra/Magick.NET : .NET wrapper for the popular ImageMagick - https://docs.microsoft.com/en-us/xamarin/xamarin-forms/user-interface/graphics/skiasharp/ : SkiaSharp Graphics in Xamarin.Forms - https://github.com/mono/SkiaSharp : .NET wrapper for Google's Skia - https://www.nuget.org/packages/CoreCompat.System.Drawing.v2/ : CoreCompat.System.Drawing.v2 - https://opencv.org/ : OpenCV Ссылка на видео: https://www.youtube.com/watch?v=S8olOE66Fnk Cлушайте все выпуски: https://anchor.fm/dotnetmore YouTube: https://www.youtube.com/playlist?list=PLbxr_aGL4q3R6kfpa7Q8biS11T56cNMf5 Обсуждайте: - VK: https://vk.com/dotnetmore - Telegram: https://t.me/dotnetmore_chat Следите за новостями: – Twitter: https://twitter.com/dotnetmore – Telegram channel: https://t.me/dotnetmore Copyright: https://creativecommons.org/licenses/by-sa/4.0/

We talk with Chris Bell, host of the Elixir Talk podcast and EMPEX conference organizer. Chris tells us about a new EMPEX chapter in the US Western states. EMPEX MTN will be in Salt Lake City, Utah. Chris started a new company called Knock using Elixir. We hear what problems it helps solve and more about his startup journey. A fun discussion with some tech insights, architecture overviews, and more on the rollercoaster of starting your own thing! Show Notes online - http://podcast.thinkingelixir.com/77 (http://podcast.thinkingelixir.com/77) Elixir Community News - https://www.twitch.tv/josevalim (https://www.twitch.tv/josevalim) – José Valim started the Advent of Code live streams on Twitch - https://github.com/josevalim/aoc/tree/main/2021 (https://github.com/josevalim/aoc/tree/main/2021) – Livebook solution source code - https://www.youtube.com/playlist?list=PLNP8vc86-SOV1ZEvXq9BLYWL586zWnF (https://www.youtube.com/playlist?list=PLNP8vc86_-SOV1ZEvX_q9BLYWL586zWnF) – Playlist of edited live streams that are shorter to watch with chapter breaks - https://twitter.com/uwucocoa/status/1467078512293130243 (https://twitter.com/_uwu_cocoa/status/1467078512293130243) – Computer vision in Livebook using OpenCV - https://github.com/cocoa-xu/evision (https://github.com/cocoa-xu/evision) – evision pulls OpenCV source code from GitHub, parse and automatically generates corresponding OpenCV-Elixir bindings. - https://twitter.com/uwucocoa/status/1466675653072371717 (https://twitter.com/_uwu_cocoa/status/1466675653072371717) – The extra steps were taken to make a reduced set of OpenCV functionality work on Nerves. - https://elixir-lang.org/blog/2021/12/03/elixir-v1-13-0-released/ (https://elixir-lang.org/blog/2021/12/03/elixir-v1-13-0-released/) – Elixir 1.13 released - blog post highlighting features - https://github.com/elixir-lang/elixir/blob/v1.13/CHANGELOG.md (https://github.com/elixir-lang/elixir/blob/v1.13/CHANGELOG.md) – The more detailed changelog that identifies the new language features. - https://twitter.com/atyborska93/status/1467520341023068162 (https://twitter.com/atyborska93/status/1467520341023068162) – Angelika Tyborska updated her Elixir Enum Cheatsheet that visually explains what functions do. - https://angelika.me/elixir-enum-cheatsheet/#slide/3 (https://angelika.me/elixir-enum-cheatsheet/#slide/3) – Angelika's update cheatsheet and the slide function - https://gleam.run/news/gleam-v0.18-released/ (https://gleam.run/news/gleam-v0.18-released/) – Gleam v0.18 released - https://blog.rentpathcode.com/introducing-eflamb%C3%A9-3065e70f9eb (https://blog.rentpathcode.com/introducing-eflamb%C3%A9-3065e70f9eb) – eFlambé by Trevor Brown - a SpawnFest winner - https://github.com/livebook-dev/livebook/blob/main/CHANGELOG.md#v040-2021-12-05 (https://github.com/livebook-dev/livebook/blob/main/CHANGELOG.md#v040-2021-12-05) – Livebook v0.4 was released - https://twitter.com/dorgan_/status/1467189972847452169 (https://twitter.com/dorgan_/status/1467189972847452169) – The Sourceror library got a new release. V0.9.0. - http://podcast.thinkingelixir.com/54 (http://podcast.thinkingelixir.com/54) – Our previous interview with Lucas San Román about Sourceror - https://github.com/supabase/supabase/tree/master/studio (https://github.com/supabase/supabase/tree/master/studio) – Supabase recently open sourced their Dashboard. - https://thinkingelixir.fireside.fm/73 (https://thinkingelixir.fireside.fm/73) – Our previous interview with Paul Copplestone at Supabase - https://www.elixirconf.eu/ (https://www.elixirconf.eu/) – ElixirConf EU 2022 is happening in London/virtual on April 6-8 Do you have some Elixir news to share? Tell us at @ThinkingElixir (https://twitter.com/ThinkingElixir) or email at show@thinkingelixir.com (mailto:show@thinkingelixir.com) Discussion Resources - https://www.empex.co/mtn (https://www.empex.co/mtn) – EMPEX MTN - Salt Lake City, UT - May 6, 2022 - https://www.youtube.com/channel/UCIYiFWyuEytDzyju6uXW40Q (https://www.youtube.com/channel/UCIYiFWyuEytDzyju6uXW40Q) – YouTube Playlists for previous EMPEX Conferences - https://knock.app/ (https://knock.app/) - https://podcasts.apple.com/us/podcast/elixir-talk/id1298287048 (https://podcasts.apple.com/us/podcast/elixir-talk/id1298287048) – Elixir Talk podcast - https://knock.app/about (https://knock.app/about) - https://www.twilio.com/ (https://www.twilio.com/) - https://sendgrid.com/ (https://sendgrid.com/) - https://www.frame.io/ (https://www.frame.io/) - https://mux.com/ (https://mux.com/) - https://twitter.com/codevisuals/status/838881724016787457?lang=en (https://twitter.com/codevisuals/status/838881724016787457?lang=en) – Flowchart for how Slack decides whether to send a notification or not. - https://github.com/sorentwo/oban (https://github.com/sorentwo/oban) Guest Information - https://twitter.com/cjbell_ (https://twitter.com/cjbell_) – on Twitter - https://github.com/cjbell/ (https://github.com/cjbell/) – on Github - https://twitter.com/empexco (https://twitter.com/empexco) – Empex on Twitter Find us online - Message the show - @ThinkingElixir (https://twitter.com/ThinkingElixir) - Email the show - show@thinkingelixir.com (mailto:show@thinkingelixir.com) - Mark Ericksen - @brainlid (https://twitter.com/brainlid) - David Bernheisel - @bernheisel (https://twitter.com/bernheisel) - Cade Ward - @cadebward (https://twitter.com/cadebward)

In this episode we look at can AI help me see better in a cost effective way! Grant Everybody, welcome to another episode of click AI radio. Okay, I have in the house today with me, someone I've been very excited to talk with. He and his organization reached out to me and I was quite surprised when I saw the cool AI solution that they have been bringing to the market. And Carlos has been giving me a little background on this. And I think you'll be excited to hear what it is he's putting together. But first and foremost, welcome, Carlos Anchia. You got Yeah. All right. There you go. Carlos, please, welcome and introduce yourself. Carlos Hey, Grant. Thanks a lot for having us on. Like you said, my name is Carlos Anchia. I'm the CEO of Plainsight AI. And we're bringing to market an end to end computer vision AI platform. I'm really, really happy to be here love talking about AI, computer vision, and how we can get more people to use it. Grant So okay, so tell me a little bit about what got you going down here. As you and I were just chatting a moment ago, there's so many components to AI, or it's such a broad range of technologies there. What got you thinking about the CV or the computer vision space? What problem? What How did you get started there? Carlos Yeah, that's a really good question. So like you said, AI, the breadth of AI is huge, you have deep learning, you have machine learning, forecasting, prediction, computer vision. And these are just a few. There's a lot of different applications for AI and places you can go down and succeed in. From our respect, we really, we really focus in on computer vision, specifically how to apply that to imagery and video. Today, there's a huge amounts of data going throughout the internet and in enterprise storage classes, where you can't really extract the value of that data unless you actually perform some sort of computer vision machine learning on that type of data. So we're really extracting the value of the picture or the video. So it can be understood by machines. So think of a dog and a cat in a in a picture, right? Those cases, the machine doesn't know it's a dog and a cat, you have to train it. And that's where computer vision comes in. And really, we got into it because we were pulled in by customers, customers of ours wanted to start doing more computer vision and leveraging our platform that we had around high throughput, ingestion, and event driven pipelines. So these customers came to us and hey, you know, this is great, we'd love to really use this for computer vision. And the more and more that kept happening, we kept retooling around the platform. And finally, the platform from end to end is purpose built to do computer vision technology. And it really allows us to focus in on on what we're good at today. Right? And that's really delivering value within the computer vision space. Grant So I remember the first time I wrote some of the OpenCV framework code, right. And that was my first introduced introduction to it. This is a number of years ago. And I started thinking, Oh, this is so cool. So I'm writing all this Python code, right, building this stuff out. And then I'm thinking, how many people you know, are actually leveraging this platform and look at even though open CV is cool, and it's got a lot of capability, it still takes a lot, you know, to get everything out of there. So can you talk about how you relate to that open CV? And what is it that you're doing relative to that? And how much easier do you guys make this? Carlos Yeah, so I mean, you hit the nail on the head there, right? So from a developer perspective, it's really around, I need to learn open CV, I need to learn Python, I need to learn containerization I need to learn deployments. There's a variety of different companies that, you know, they're all great in their own right, right. Every one of those companies that we just talked about organizations are contributing tremendously to AI. But from a developer's perspective, you really have to learn a little bit of everything to be able to orchestrate a solution. And finally, when you get to, hey, I use AI. Let's pretend we're looking at strawberries. Hey, look, I built a model that the Texas strawberry that is your over the moon excited, but the very next thing is around, okay, how do I take that and deploy it 1000 times over in a field across the world and understand how to make that in an operational fashion where you know, it can be supported, maintained update, and that's really where we have this this crux of an organization where it's really different building something on a on a desk for a one time use. And and there's a lot of wins through that process. But then taking that and operationalizing for business driving revenues driving corporate goals around, why would that feature is being implemented, that's really where we come in, we want to be able to take off that that single single path of workflow where it's a little bit of everything to orchestrate a solution, and provide a centralized place where other people, including developers can go and help build that workflow in a meaningful way where it's complete. Grant So operationalizing, those models, I find, that's one of the biggest, or the most challenging aspects to this, it's one thing as you know, to, to build out enough to sort of prove something out and get some of the initial feedback, but to actually get it into production. I think I saw MIT not long ago, maybe this a year ago, now, they had come out with this report, it was through the Boston Consulting Group as well, they'd mentioned something about, hey, you know, 10% of organizations doing AI are getting return on their investment. And, and, of course, when you look at all of the investment of the takes for the business to really stand up all the data scientists and all the ML work. And you can see why the numbers translate that way. So to me, it feels like not only doing this in the area of CV, but the problem you're really trying to solve it feels like is you're attacking that ROI problem, which is you could take this kind of capability into business say you don't have to stand up all of these deep technical capabilities. Rather, you can achieve ROI sooner than rather than laters. Is that Is that accurate? Carlos That's correct. And I think it's really through the adoption of technology and you hit a you hit a really strong point for us there around the the difference between it works and it's operational. That's that's really the path of your your, you're less there in the CV world and more there with the DevOps ml ops portion of it, getting machines running consistently, with the right versions of deployment strategy, that latter half of it, just as important as the model building pieces of it. But even after you get to that piece, you need a way to improve, and improvement in the model is very costly if it's not automated. Because I mean, you can just look at the the loss for a simple detector, like a strawberry, where, you know, if the model starts to perform poorly, you're not pulling as many strawberries out of the field. So you need a way to be able to update that model, quickly, get training data into a platform and push a new model back out. And it's really around how fast you can go end to end with that workflow. Again, and again. And again. And again, this is that continuous improvement that we have born into us from previous software development life, but really in in machine learning and computer vision, your ability to train, retrain, and redeploy. That is where you really get the benefit out of your workflow. Grant Well, that really confirms my experience with AI will. Typically I'll refer to the term that we've been using, as we call it a SmartStep. It's that notion that I need to be able to refactor my models and take in consideration that changed context around me, whether it comes in from the world or from the customer, or whatever that means, some level of adjustments taken place that begins to invalidate my previous AI model. And I need to be able to quickly make those adjustments. That's fascinating. How long typically does it take for you to do that kind of refactoring of your models? Is that Is it a day? Is it a year, a month? Or the answer is? Well, it depends. Carlos So it's twofold, right? So it's, it's hours to do that. But it really depends on the complexity of the model, and how long you have to train. But in an automated workflow, you're you're continuously adding data to your training set, that are lower quality predictions, where you can retrain automatically when you hit a certain threshold, and then validate the model and push that back out into your production alized environment. So it's it when you go to develop these sort of workflows. You really have to start with whatever I build, I know I have to improve on later. So that improvement cycle ends up costing a lot if it's not part of the initial discussion around how do we count strawberries, right? So it's evident you and I can nerd out on this. Let me shift the focus a little bit and ask about it from say your your customer Write from their perspective, what is it that they need to be able to do to be successful with your solution? What skills or capabilities do they have to bring to the table? Yeah, and I think it's, I think I have this conversation a lot with our clients. And it's really less about them having technology around data science and building model, and more around a collaborative environment, where organizations, you know, they have a culture of success. But that culture of success is really borne by holding hands through the fire, it's, it's being able to commit and lean in when the organization sees something that's really important to them, either either from a technical perspective or revenue perspective. And it's these companies and these types of people that get to rally around a centralized platform where they can build and collaborate with machine learning computer vision applications. And, you know, it's it's a, it's really interesting to see the companies that succeed here, because it's really based on a culture of winning, right, where the wind doesn't have to be the hardest, most technical, logically difficult problem, because complexity really drives timelines. And if you're looking to change from an organization's perspective, start getting the little wins, get the little wins, start having some adoption within the company around, wow, computer vision is working. We've identified these problems in a few hours, we have a solution deployed, you start building this sense of confidence in the organization where you can take on those larger tasks. But you have to start with a build up, you can't just go right to the highest ROI problem. No one starts at human genome sequencing. Grant Have you, or do we got a problem? Yeah. Back up, back up. So So all right. So it means to me it sounds like as an organization to succeed with this getting my problem definition, understood or crisply put together first, what would be an obvious thing to do? But how long does it take for me to iterate? Before I know that I've got value, that I've pursued the right level of the problem? You You made an interesting comment a minute ago, you're like, oh, within a couple hours, I could potentially retrain the model and have that back operationally. That means if I can fail fast, right, if I can pick my problem space, get something out there operation, try it fail fast, and then continue to iterate with AI as my helper that that's really, really quite powerful is that the model that the your person? Carlos It is the model? That's exactly right. And it's not just hours to retrain, it takes hours to start, right. And just to highlight you kind of started with, we have to define that problem set first. So even after we define that problem set, a lot of times we have to go back and redefine that problem set, and really the piece around failing fast. It's it's experimentation. And do we have the right cameras? Do we have the right vantage point is the model correct, you want to be able to cycle as fast as you can through that experimentation phase. And sometimes you have to go back and redefine that problem set. Because you're learning more as you go, right? And you're evolving into okay, I now understand the corner cases a bit better. And with the platform, you really can cycle that quickly. I mean, machine learning at scale is really how fast you can iterate through improvement. Grant It's quite, I think it's quite a testament to how the AI just world in general is improving. I know that you and I were talking earlier, years ago, you know, when I first started writing some TensorFlow code and Keras code, you know, the, the time it took to fail was much longer, right. And then the cycles were huge and, and getting this down to a matter of hours or even a few days, you know, for an enterprise's is massive. What's the, from what you've seen terms of different industries? Are there certain industries that tend to be leaning into this and adopting it or the is there no pattern yet? Carlos No, there's a there's a definite pattern. And in 2022, we'll all see kind of what that what that's looking like. And it's really an industry that has traditionally not been able to go through that digital transformation. So think of think of think of a piece that's a very manual piece, right, like physical inspection, where humans would look at something, they'd write down their notes on a piece of paper, then that that item would go through either pass or fail or some criteria for rework. That's all possible. Now with computer vision years ago, that was impossible the accuracy wasn't high enough, plain and simple. It just took it wasn't, it wasn't better than a human. Now we have models that are better than a human for visual inspection. And these industries are digitizing their workflow. So it's not only the feature of computer vision, but it's also now I have a digital record of all the transactions, I have extracted video information, it makes auditing easy for those sectors that have a lot of regulatory compliance, those that require proactive compliance to audit requirements, as well as visibility. Visibility has always been an it's funny when we talk about visibility, but it's computer vision, but like a lot of human processes, that there's zero visibility in it there, it's really difficult to audit, you know, why is this working better or not? So having that, that digitization of the flow with the feature of computer vision allows us to extract the value. And industries like agriculture are going I mean, agriculture has been a leader in technology for a while, but now you're really seeing adoption at livestock and row crop with drone technology. It's a very rich image environment, medical space, medical space for computer vision in 2022 to 2028, is estimated to be billions of dollars just with medical imaging. And that's not that's not the the total addressable market for the hardware, it's just the imaging piece. So we see we see a lot of growth in sectors that are going through this digital transformation that are adopting technologies that are now getting to the point where they can get pushed down into the masses instead of just the top five companies in the world. Grant Excellent, it seems to you part of your comment earlier made me think about process optimization for organizations, and the ability to extract processes, you're familiar with process mining, right? The ability to extract, you know, out of logs of these organizations and doing something like that where you can produce this visual representation of that, and then building models against that, to optimize your process might be an interesting use case. Yeah, that's fascinating. Carlos That's a really good point, right? Because that's, that's a different portion of AI that can be applied to like just log analysis, that then would allow you to go back and Okay, now that we have the process mind, where can we improve along the process? Grant Yeah, yeah. Amazing. So many uses and use cases around around this CV area, for sure. So let's say that someone listening to this wanted to learn more about it, where would they go? How would they? How would they find more about your organization? Carlos You can find us everywhere, right? We have a website, plainsight.ai. We're all over LinkedIn, we have Twitter, we're on Reddit, we have a Medium blog, there's a Slack channel where we geek out around computer vision use cases and how we can improve the world through computer vision. We're really we're really out there and feel free if you have questions come reach out to us. We have amazing staff that are looking to empower people in AI. So if it's through just just a question around how does this thing work, we'd love to talk to you if it's Hey, we're kind of stuck in our journey. We need some help reach out to us we can help you. Grant That's awesome. Carlos I can't thank you enough for reaching out to me and for a listening to click AI radio, but also for reaching out and sharing what it is you are you and your organization are bringing to the market think you're solving some awesome problems. Carlos Thanks a lot, Grant. Appreciate it. always appreciated talking about computer vision and AI and thank you to you and your listeners and really appreciate what you're doing to the AI space. Grant Alright, thanks again, Carlos. And again, everybody. Thanks for joining and until next time, go get some computer vision from Plainsight.

In this episode we look at can AI help me see better in a cost effective way! Grant Everybody, welcome to another episode of click AI radio. Okay, I have in the house today with me, someone I've been very excited to talk with. He and his organization reached out to me and I was quite surprised when I saw the cool AI solution that they have been bringing to the market. And Carlos has been giving me a little background on this. And I think you'll be excited to hear what it is he's putting together. But first and foremost, welcome, Carlos Anchia. You got Yeah. All right. There you go. Carlos, please, welcome and introduce yourself. Carlos Hey, Grant. Thanks a lot for having us on. Like you said, my name is Carlos Anchia. I'm the CEO of Plainsight AI. And we're bringing to market an end to end computer vision AI platform. I'm really, really happy to be here love talking about AI, computer vision, and how we can get more people to use it. Grant So okay, so tell me a little bit about what got you going down here. As you and I were just chatting a moment ago, there's so many components to AI, or it's such a broad range of technologies there. What got you thinking about the CV or the computer vision space? What problem? What How did you get started there? Carlos Yeah, that's a really good question. So like you said, AI, the breadth of AI is huge, you have deep learning, you have machine learning, forecasting, prediction, computer vision. And these are just a few. There's a lot of different applications for AI and places you can go down and succeed in. From our respect, we really, we really focus in on computer vision, specifically how to apply that to imagery and video. Today, there's a huge amounts of data going throughout the internet and in enterprise storage classes, where you can't really extract the value of that data unless you actually perform some sort of computer vision machine learning on that type of data. So we're really extracting the value of the picture or the video. So it can be understood by machines. So think of a dog and a cat in a in a picture, right? Those cases, the machine doesn't know it's a dog and a cat, you have to train it. And that's where computer vision comes in. And really, we got into it because we were pulled in by customers, customers of ours wanted to start doing more computer vision and leveraging our platform that we had around high throughput, ingestion, and event driven pipelines. So these customers came to us and hey, you know, this is great, we'd love to really use this for computer vision. And the more and more that kept happening, we kept retooling around the platform. And finally, the platform from end to end is purpose built to do computer vision technology. And it really allows us to focus in on on what we're good at today. Right? And that's really delivering value within the computer vision space. Grant So I remember the first time I wrote some of the OpenCV framework code, right. And that was my first introduced introduction to it. This is a number of years ago. And I started thinking, Oh, this is so cool. So I'm writing all this Python code, right, building this stuff out. And then I'm thinking, how many people you know, are actually leveraging this platform and look at even though open CV is cool, and it's got a lot of capability, it still takes a lot, you know, to get everything out of there. So can you talk about how you relate to that open CV? And what is it that you're doing relative to that? And how much easier do you guys make this? Carlos Yeah, so I mean, you hit the nail on the head there, right? So from a developer perspective, it's really around, I need to learn open CV, I need to learn Python, I need to learn containerization I need to learn deployments. There's a variety of different companies that, you know, they're all great in their own right, right. Every one of those companies that we just talked about organizations are contributing tremendously to AI. But from a developer's perspective, you really have to learn a little bit of everything to be able to orchestrate a solution. And finally, when you get to, hey, I use AI. Let's pretend we're looking at strawberries. Hey, look, I built a model that the Texas strawberry that is your over the moon excited, but the very next thing is around, okay, how do I take that and deploy it 1000 times over in a field across the world and understand how to make that in an operational fashion where you know, it can be supported, maintained update, and that's really where we have this this crux of an organization where it's really different building something on a on a desk for a one time use. And and there's a lot of wins through that process. But then taking that and operationalizing for business driving revenues driving corporate goals around, why would that feature is being implemented, that's really where we come in, we want to be able to take off that that single single path of workflow where it's a little bit of everything to orchestrate a solution, and provide a centralized place where other people, including developers can go and help build that workflow in a meaningful way where it's complete. Grant So operationalizing, those models, I find, that's one of the biggest, or the most challenging aspects to this, it's one thing as you know, to, to build out enough to sort of prove something out and get some of the initial feedback, but to actually get it into production. I think I saw MIT not long ago, maybe this a year ago, now, they had come out with this report, it was through the Boston Consulting Group as well, they'd mentioned something about, hey, you know, 10% of organizations doing AI are getting return on their investment. And, and, of course, when you look at all of the investment of the takes for the business to really stand up all the data scientists and all the ML work. And you can see why the numbers translate that way. So to me, it feels like not only doing this in the area of CV, but the problem you're really trying to solve it feels like is you're attacking that ROI problem, which is you could take this kind of capability into business say you don't have to stand up all of these deep technical capabilities. Rather, you can achieve ROI sooner than rather than laters. Is that Is that accurate? Carlos That's correct. And I think it's really through the adoption of technology and you hit a you hit a really strong point for us there around the the difference between it works and it's operational. That's that's really the path of your your, you're less there in the CV world and more there with the DevOps ml ops portion of it, getting machines running consistently, with the right versions of deployment strategy, that latter half of it, just as important as the model building pieces of it. But even after you get to that piece, you need a way to improve, and improvement in the model is very costly if it's not automated. Because I mean, you can just look at the the loss for a simple detector, like a strawberry, where, you know, if the model starts to perform poorly, you're not pulling as many strawberries out of the field. So you need a way to be able to update that model, quickly, get training data into a platform and push a new model back out. And it's really around how fast you can go end to end with that workflow. Again, and again. And again. And again, this is that continuous improvement that we have born into us from previous software development life, but really in in machine learning and computer vision, your ability to train, retrain, and redeploy. That is where you really get the benefit out of your workflow. Grant Well, that really confirms my experience with AI will. Typically I'll refer to the term that we've been using, as we call it a SmartStep. It's that notion that I need to be able to refactor my models and take in consideration that changed context around me, whether it comes in from the world or from the customer, or whatever that means, some level of adjustments taken place that begins to invalidate my previous AI model. And I need to be able to quickly make those adjustments. That's fascinating. How long typically does it take for you to do that kind of refactoring of your models? Is that Is it a day? Is it a year, a month? Or the answer is? Well, it depends. Carlos So it's twofold, right? So it's, it's hours to do that. But it really depends on the complexity of the model, and how long you have to train. But in an automated workflow, you're you're continuously adding data to your training set, that are lower quality predictions, where you can retrain automatically when you hit a certain threshold, and then validate the model and push that back out into your production alized environment. So it's it when you go to develop these sort of workflows. You really have to start with whatever I build, I know I have to improve on later. So that improvement cycle ends up costing a lot if it's not part of the initial discussion around how do we count strawberries, right? So it's evident you and I can nerd out on this. Let me shift the focus a little bit and ask about it from say your your customer Write from their perspective, what is it that they need to be able to do to be successful with your solution? What skills or capabilities do they have to bring to the table? Yeah, and I think it's, I think I have this conversation a lot with our clients. And it's really less about them having technology around data science and building model, and more around a collaborative environment, where organizations, you know, they have a culture of success. But that culture of success is really borne by holding hands through the fire, it's, it's being able to commit and lean in when the organization sees something that's really important to them, either either from a technical perspective or revenue perspective. And it's these companies and these types of people that get to rally around a centralized platform where they can build and collaborate with machine learning computer vision applications. And, you know, it's it's a, it's really interesting to see the companies that succeed here, because it's really based on a culture of winning, right, where the wind doesn't have to be the hardest, most technical, logically difficult problem, because complexity really drives timelines. And if you're looking to change from an organization's perspective, start getting the little wins, get the little wins, start having some adoption within the company around, wow, computer vision is working. We've identified these problems in a few hours, we have a solution deployed, you start building this sense of confidence in the organization where you can take on those larger tasks. But you have to start with a build up, you can't just go right to the highest ROI problem. No one starts at human genome sequencing. Grant Have you, or do we got a problem? Yeah. Back up, back up. So So all right. So it means to me it sounds like as an organization to succeed with this getting my problem definition, understood or crisply put together first, what would be an obvious thing to do? But how long does it take for me to iterate? Before I know that I've got value, that I've pursued the right level of the problem? You You made an interesting comment a minute ago, you're like, oh, within a couple hours, I could potentially retrain the model and have that back operationally. That means if I can fail fast, right, if I can pick my problem space, get something out there operation, try it fail fast, and then continue to iterate with AI as my helper that that's really, really quite powerful is that the model that the your person? Carlos It is the model? That's exactly right. And it's not just hours to retrain, it takes hours to start, right. And just to highlight you kind of started with, we have to define that problem set first. So even after we define that problem set, a lot of times we have to go back and redefine that problem set, and really the piece around failing fast. It's it's experimentation. And do we have the right cameras? Do we have the right vantage point is the model correct, you want to be able to cycle as fast as you can through that experimentation phase. And sometimes you have to go back and redefine that problem set. Because you're learning more as you go, right? And you're evolving into okay, I now understand the corner cases a bit better. And with the platform, you really can cycle that quickly. I mean, machine learning at scale is really how fast you can iterate through improvement. Grant It's quite, I think it's quite a testament to how the AI just world in general is improving. I know that you and I were talking earlier, years ago, you know, when I first started writing some TensorFlow code and Keras code, you know, the, the time it took to fail was much longer, right. And then the cycles were huge and, and getting this down to a matter of hours or even a few days, you know, for an enterprise's is massive. What's the, from what you've seen terms of different industries? Are there certain industries that tend to be leaning into this and adopting it or the is there no pattern yet? Carlos No, there's a there's a definite pattern. And in 2022, we'll all see kind of what that what that's looking like. And it's really an industry that has traditionally not been able to go through that digital transformation. So think of think of think of a piece that's a very manual piece, right, like physical inspection, where humans would look at something, they'd write down their notes on a piece of paper, then that that item would go through either pass or fail or some criteria for rework. That's all possible. Now with computer vision years ago, that was impossible the accuracy wasn't high enough, plain and simple. It just took it wasn't, it wasn't better than a human. Now we have models that are better than a human for visual inspection. And these industries are digitizing their workflow. So it's not only the feature of computer vision, but it's also now I have a digital record of all the transactions, I have extracted video information, it makes auditing easy for those sectors that have a lot of regulatory compliance, those that require proactive compliance to audit requirements, as well as visibility. Visibility has always been an it's funny when we talk about visibility, but it's computer vision, but like a lot of human processes, that there's zero visibility in it there, it's really difficult to audit, you know, why is this working better or not? So having that, that digitization of the flow with the feature of computer vision allows us to extract the value. And industries like agriculture are going I mean, agriculture has been a leader in technology for a while, but now you're really seeing adoption at livestock and row crop with drone technology. It's a very rich image environment, medical space, medical space for computer vision in 2022 to 2028, is estimated to be billions of dollars just with medical imaging. And that's not that's not the the total addressable market for the hardware, it's just the imaging piece. So we see we see a lot of growth in sectors that are going through this digital transformation that are adopting technologies that are now getting to the point where they can get pushed down into the masses instead of just the top five companies in the world. Grant Excellent, it seems to you part of your comment earlier made me think about process optimization for organizations, and the ability to extract processes, you're familiar with process mining, right? The ability to extract, you know, out of logs of these organizations and doing something like that where you can produce this visual representation of that, and then building models against that, to optimize your process might be an interesting use case. Yeah, that's fascinating. Carlos That's a really good point, right? Because that's, that's a different portion of AI that can be applied to like just log analysis, that then would allow you to go back and Okay, now that we have the process mind, where can we improve along the process? Grant Yeah, yeah. Amazing. So many uses and use cases around around this CV area, for sure. So let's say that someone listening to this wanted to learn more about it, where would they go? How would they? How would they find more about your organization? Carlos You can find us everywhere, right? We have a website, plainsight.ai. We're all over LinkedIn, we have Twitter, we're on Reddit, we have a Medium blog, there's a Slack channel where we geek out around computer vision use cases and how we can improve the world through computer vision. We're really we're really out there and feel free if you have questions come reach out to us. We have amazing staff that are looking to empower people in AI. So if it's through just just a question around how does this thing work, we'd love to talk to you if it's Hey, we're kind of stuck in our journey. We need some help reach out to us we can help you. Grant That's awesome. Carlos I can't thank you enough for reaching out to me and for a listening to click AI radio, but also for reaching out and sharing what it is you are you and your organization are bringing to the market think you're solving some awesome problems. Carlos Thanks a lot, Grant. Appreciate it. always appreciated talking about computer vision and AI and thank you to you and your listeners and really appreciate what you're doing to the AI space. Grant Alright, thanks again, Carlos. And again, everybody. Thanks for joining and until next time, go get some computer vision from Plainsight.

On this week's show, Seth welcomes grand prize winners of the 2021 OpenCV AI Competition. Satya Mallick from OpenCV is here with Ye Lu and Him Wai (Michael) Ng from Team Cortic Tigers, who will demo their award-winning project - Cortic Edge Platform (CEP) which aims to democratize AI for everyone!Jump to:[00:17] Welcome to the AI Show[01:01] What is the OpenCV AI Competition[03:15] 2021 Grand Prize Winner - Cortic Edge Platform (CEP) - allowing beginners and advanced programmers to start doing rapid AI prototyping[05:21] CEP use cases[10:01] Using Microsoft MakeCode to build CEP[13:34] What's next from OpenCV AI Learn more:OpenCV AI Competition https://opencv.org/opencv-ai-competition-2021/Cortic Technology GitHub repo https://github.com/cortictechnology/cepShop the OAK-D camera https://shop.luxonis.com/products/1098obcenclosure Kickstarter campaign https://www.opencv.org/kickstarterZero to Hero Machine Learning on Azure https://aka.ms/ZerotoHero/MLonAzureZero to Hero Azure AI https://aka.ms/ZerotoHero/AzureAICreate a Free account (Azure) https://aka.ms/aishow-seth-azurefreeFollow Seth https://twitter.com/sethjuarezFollow Satya https://twitter.com/LearnOpenCVFollow Cortic Technology https://twitter.com/CorticTechnolo1Follow Cortic Technology Group https://www.linkedin.com/company/cortic/Don't miss new episodes, subscribe to the AI Show https://aka.ms/AIShowsubscribeAI Show Playlist https://aka.ms/AIShowPlaylistJoin us every other Friday, for an AI Show livestream on Learn TV and YouTube https://aka.ms/LearnTV - https://aka.ms/AIShowLive

On this week's show, Seth welcomes grand prize winners of the 2021 OpenCV AI Competition. Satya Mallick from OpenCV is here with Ye Lu and Him Wai (Michael) Ng from Team Cortic Tigers, who will demo their award-winning project - Cortic Edge Platform (CEP) which aims to democratize AI for everyone!Jump to:[00:17] Welcome to the AI Show[01:01] What is the OpenCV AI Competition[03:15] 2021 Grand Prize Winner - Cortic Edge Platform (CEP) - allowing beginners and advanced programmers to start doing rapid AI prototyping[05:21] CEP use cases[10:01] Using Microsoft MakeCode to build CEP[13:34] What's next from OpenCV AI Learn more:OpenCV AI Competition https://opencv.org/opencv-ai-competition-2021/Cortic Technology GitHub repo https://github.com/cortictechnology/cepShop the OAK-D camera https://shop.luxonis.com/products/1098obcenclosure Kickstarter campaign https://www.opencv.org/kickstarterZero to Hero Machine Learning on Azure https://aka.ms/ZerotoHero/MLonAzureZero to Hero Azure AI https://aka.ms/ZerotoHero/AzureAICreate a Free account (Azure) https://aka.ms/aishow-seth-azurefreeFollow Seth https://twitter.com/sethjuarezFollow Satya https://twitter.com/LearnOpenCVFollow Cortic Technology https://twitter.com/CorticTechnolo1Follow Cortic Technology Group https://www.linkedin.com/company/cortic/Don't miss new episodes, subscribe to the AI Show https://aka.ms/AIShowsubscribeAI Show Playlist https://aka.ms/AIShowPlaylistJoin us every other Friday, for an AI Show livestream on Learn TV and YouTube https://aka.ms/LearnTV - https://aka.ms/AIShowLive

Computer Vision is a popular field of Artificial Intelligence that enables applications such as facial recognition and self-driving cars. In this episode we hear from Dr. Satya Mallick who has been working in the field since it's birth on all the various aspect of Computer Vision and how you can learn it to break into the field of AI. How to Connect with Satya: - COMPANY WEBSITE: https://learnopencv.com/ - TWITTER: https://twitter.com/learnopencv - LINKEDIN: https://www.linkedin.com/in/satyamallick - EMAIL: spmalek@opencv.com Links referenced in episode: https://www.image-net.org/challenges/LSVRC/ https://aipaygrad.es/ Learn data skills at our academy and elevate your career. Start for free at https://ftdacademy.com/pod --- Send in a voice message: https://anchor.fm/ftdacademy/message

02:01 - Sy's Superpower: Making Complex Topics Digestible * Sy on YouTube: "Computer Science Explained with my Cats" (https://www.youtube.com/SyBrandPlusCats) 06:28 - Approaching Learning to Code: Do Something That Motivates You * Greater Than Code Episode 246: Digital Democracy and Indigenous Storytelling with Rudo Kemper (https://www.greaterthancode.com/digital-democracy-and-indigenous-storytelling) * Ruby For Good (https://rubyforgood.org/) * Terrastories (https://terrastories.io/) 11:25 - Computers Can Hurt Our Bodies! * Logitech M570 Max (https://www.amazon.com/Logitech-M570-Wireless-Trackball-Mouse/dp/B0043T7FXE) * Dvorak Keyboard (https://www.dvorak-keyboard.com/) 13:57 - Motivation (Cont'd) * Weekend Game Jams * The I Do, We Do, You Do Pattern (https://theowlteacher.com/examples-of-i-do-you-do-we-do/) 22:15 - Sy's Content (Cont'd) * Sy on YouTube: "Computer Science Explained with my Cats" (https://www.youtube.com/SyBrandPlusCats) * Content Creation and Choosing Topics 33:58 - Code As Art * code:art (https://code-art.xyz/) / @codeart_journal (https://twitter.com/codeart_journal) * trashheap (https://trashheap.party/) / @trashheapzine (https://twitter.com/trashheapzine) * Submission Guidelines (https://trashheap.party/submit/) * Casey's Viral TikTok! (https://www.tiktok.com/@heycaseywattsup/video/6988571925811367173?lang=en&is_copy_url=1&is_from_webapp=v1) 41:34 - #include ＜C++＞ (https://www.includecpp.org/) * Lessons learned creating an inclusive space in a decades old community (Sy's Talk) (https://developerrelations.com/community/lessons-learned-creating-an-inclusive-space-in-a-decades-old-community) * QueerJS (https://queerjs.com/) * Emscripten (https://emscripten.org/) * Graphiz it! (http://graphviz.it/#/gallery) Reflections: Mandy: Digging into Sy's videos. Casey: Working within content creation constraints. Sy: Make a video on register allocation. This episode was brought to you by @therubyrep (https://twitter.com/therubyrep) of DevReps, LLC (http://www.devreps.com/). To pledge your support and to join our awesome Slack community, visit patreon.com/greaterthancode (https://www.patreon.com/greaterthancode) To make a one-time donation so that we can continue to bring you more content and transcripts like this, please do so at paypal.me/devreps (https://www.paypal.me/devreps). You will also get an invitation to our Slack community this way as well. Transcript: Software is broken, but it can be fixed. Test Double's superpower is improving how the world builds software by building both great software and great teams and you can help. Test Double is looking for empathetic senior software engineers and dev ops engineers. We work in JavaScript, Ruby, Elixir, and a lot more. Test Double trusts developers with autonomy and flexibility at a 100% remote employee-owned software consulting agency. Are you trying to grow? Looking for more challenges? Enjoy lots of variety in projects working with the best teams in tech as a developer consultant at Test Double. Find out more and check out remote openings at link.testdouble.com/join. That's link.testdouble.com/join. MANDY: Hello and welcome to Greater Than Code, Episode 247. My name is Mandy Moore and I'm here with my friend, Casey Watts. CASEY: Hi, I'm Casey, and we're both here with our guest today, Sy Brand. SY: Hey, everyone! CASEY: Sy is Microsoft's C++ Developer Advocate. Their background is in compilers and debuggers for embedded accelerators. They're particularly interested in generic library design, making complex concepts understandable, and making our communities more welcoming and inclusive. They can usually be found on Twitter, playing with their three cats, writing, or watching experimental movies. Hi, Sy! Good to have you. SY: Hey, thanks for having me on. CASEY: The first question we like to ask, I think you're prepared for it, is what is your superpower and how did you acquire it? SY: Yeah, so very topically, I think one of my superpowers is forgetting what topics I want to talk about when recording podcasts and that, I acquired through having ADHD and forgetting to write things down. But I did write things down this time so maybe that won't be too much of a problem. But I think one of my other ones is making complex topics digestible, trying to take computer science topics and distill them down into things which are understandable without necessarily having a lot of the background knowledge, the resources you'd expect. I gained that mostly through my background in computer science and then my interest in public speaking and communication and performance poetry, trying to blend those together to make things easier to understand, lower the barrier for entry. CASEY: I love it. Making complex topics digestible. That's definitely a skill we need more of in the world. MANDY: Absolutely. So Casey told me you are a bit of a teacher and you do a lot of teaching on, is it YouTube? So making things easier to digest. Like I said, during the preshow, I've been trying to learn to code on and off for 12 years, as long as I've had this career, and I've started and stopped, gotten frustrated and stopped, and I've tried different things. I've had mentors and I feel like I've let my mentors down and I've tried this and that. I've tried the code academy and I don't know. So how do you do it? Can you tell us a little bit about how you do that? SY: Sure. So most of the topics that I am interested in teaching is, because I come from a background of compilers and debuggers and very low-level systems, those are the things that I want people to get excited about because I think people look at compilers, or C++, or low-level programming and think, “Oh, this is not very interesting,” or new, or it's too complex, or it requires too much of a degree, or whatever. But none of that is true. You can write a compiler without having to have a lot of the background knowledge you might expect and you can learn C++ without having to – it can be a lot easier than people make art. So I want to make these concepts seem interesting and understandable because they're deeply interesting to me and they've been working on them for a large part of my life and I still love it and find them fascinating. So I want to share that with people. CASEY: What's your motivation when you're working on these? Is it to understand things that are complex, or are you solving problems you have, or other people have, or maybe a blend, or other motivations? I'm wondering what gets you so pumped about it. SY: Yeah, so I think it's a few different things. I make videos on Twitter, or YouTube, things like that of explaining concepts that I'm already familiar with and it's pretty much stuff that I could write an entire video off the top of my head without having to do any research. So I've done videos on explaining what a compiler is and all the stages of compilation, or a video on higher cash performance works, or [in audible 05:48] cash configurancy, garbage collection. These are all things I could just sit down and write something on and don't have to do a lot of research. Then there's the more exploratory stuff. I've been live streaming the development of a Ranges library for C++, which is being able to compose operations, building up a pipeline of operations for your data and then declarative manner so that you don't have to deal with a lot of memory allocations and moving data, or a range yourself. You just say, “Here's all the steps that I want to occur,” and then someone who has written all of these pipeline operations deals with how that actually happens. I've been developing that library live and trying to teach myself hired to do all of these things as while also teaching other people at the same time. MANDY: So is it right to assume that maybe I've been going about learning to code in all the wrong ways and that I've just picked a language and tried to dive in, or did I miss some of the conceptual stuff? And if so, as I suspect, a lot of the conceptual stuff has gone over my head. So where do you suggest, if you were giving me advice, which yes, you are giving me advice. [laughter] Where would you suggest, as a brand-new beginner coder, what kind of software concepts I need to research and understand before actually diving into an actual programming language? SY: Honestly, I don't think that there's a single answer there and I don't think there's a lot of wrong answers there. From my perspective, the best way to learn how to code is doing something that motivates you and that gets you excited because coding is hard and when you hit those bumps and things are going wrong, if you don't have that motivation to keep going, then it's very easy to stop. I know I've done it in trying to learn certain concepts and things like that before, because I felt like, “Oh, I should learn this thing, but I wasn't really interested in it,” and then I find out it was hard and stopped. The best way that I learn is finding something where I'm like, “Hey, I want to build this thing,” or “I want to understand this because I want to solve this problem,” or “because I want to dove on that knowledge with something else.” It's always the motivation, but then I'm coming from if you're someone with ADHD, or something like me, then it's pretty much impossible to do anything without [chuckles] having a strong motivation behind it. So that kind of comes into my way of learning as well. MANDY: That's super interesting. Actually, the last episode we did was with Rudo Kemper and he did a project with Ruby for Good. I went to that and I actually got really excited, intrigued, and wanted to get involved and learn how to code because I was really interested and passionate about the project that he presented, which was Terrastories, which was handing down indigenous knowledge technologically so that stories aren't lost in just having oral traditions, that these stories are actually being recorded and are living somewhere on the internet. So that's really interesting. I went to that and then of course, pandemic happened. It didn't happen again last year, but I'm thinking about going back this year. I'm hoping maybe I can be on a team with somebody that could just shadow and sit there and maybe Casey would let me be that person because rumor has it, Casey is going to be there. Ruby for Good on the East Coast in the fall. CASEY: Yeah, I'll be there. I'd be happy to have you shadow me. Also, my role lately has been a higher level. Last time I was a product manager for the team not coding and this year I'm going to be helping the teams be happy and effective across the board because there's always a team, or two that need some alignment work so that they can be productive the whole weekend. MANDY: That's interesting. Okay. Well, I'm sure I'll find somebody who wouldn't mind me doing a kind of shadow. CASEY: For sure. MANDY: Yeah, cool. CASEY: That's the kind of environment it is. MANDY: Absolutely. CASEY: Yeah. SY: That definitely sounds like the right kind of thing like something where you hear about something, or you look at this project and you think, “Hey, I want to get involved. I want to contribute to this.” That's what can drive a positive learning experience, I think it's that motivation and that motivation could just be, “Hey, I want to get into the tech industry because it pays well and we need money to live because capitalism.” That's like totally legit as well. Whatever you find motivates you to work. MANDY: Yeah, that's why I'm here. I had to find a way for my daughter and I to live. SY: Yeah. MANDY: So I got into tech and podcasts and then I'm working for all these people who I always considered so much smarter than me. I was like, “I could never learn that. I'm not good enough.” But now since joining the podcast as a host and coming on here, I'm feeling more and more like I am smart enough, I could do the thing and so, I'm actually really getting into it more. But it's just that being on the computer for so many hours doing the work stuff makes it hard to also break into the wanting to do the learning outside of my work hours – [overtalk] SY: Right, yeah. MANDY: Because it's so much computering. SY: Yeah, or just split the good screen from bad screen. CASEY: I've been computering so much, I have a tendonitis in my right pinky now from using the arrow keys on the keyboard too much, I think and bad posture, which I've been working on for years. Computers can hurt our bodies. SY: Yeah, definitely. I use the Logitech M570 mouse, which I switched to a number of years ago and was one of the best changes I ever made for using the computer and also, switching to Dvorak for keyboard layout. CASEY: Okay. I use that, too. SY: Nice! CASEY: Dvorak. It's not better, but I learned it. [laughter] It might be more better for my health maybe, but I'm not faster. That's what people always ask. SY: I'm definitely – [overtalk] CASEY: Instead of ASDF, it's a AOEU under your fingers; the common letters right at your fingertips. You don't need the semicolon under your right pinky. [laughter] Why is that there? SY: Yeah. MANDY: Yeah. I was going to ask for us what you were even talking about there. So it's just basically reconfiguring your keyboard to not be QWERTY thing? SY: Yeah, exactly. MANDY: Okay. SY: That means you have to completely relearn how to type, which can take a while. Like when I completely stopped using QWERTY at all and just switched to Dvorak, I didn't even buy a Dvorak keyboard, I just printed out the keyboard layout and stuck it to my monitor and just learned. For the first while, it's excruciating because you're trying to type an email and you're typing 15 words per minute, or something. That's bad. I did definitely did get faster shifting to Dvorak. Before I think I used to type at like 70, 80; I type around a 100 words per minute so it changed my speed a bit. But to be fair, I don't think I typed properly on QWERTY. I switched 10 years ago, though so I can't even remember a whole lot. [chuckles] MANDY: That's interesting, though. That gives me something I want to play around with right there and it's not even really coding. [laughter] It's just I'll be just trying to teach myself to type in a different way. That's really interesting. Thank you. [chuckles] CASEY: Yeah. It was fun for when I learned it, too. I think I learned in middle school and I was I practiced on AIM, AOL Instant Messenger, and RuneScape. SY: Nice. CASEY: I didn't dare practice while I had essays due and I had to write those up. That was too stressful. [laughter] CASEY: Summer was better for me. SY: Yeah, I switched during a summer break at university. CASEY: Low stakes. I needed the low stakes for that to succeed. SY: [laughs] Yeah. CASEY: We were talking about what motivates you to learn programming and I wrote up a story about that for me actually recently. SY: Okay. CASEY: At the highest level, my first programming class, we modeled buoys and boats and it was so boring. I don't know why we were doing it. It didn't have a purpose. There was no end goal, no user, nobody was ever going to use the code. It was fine for learning concepts, I guess, but it wasn't motivated and I hated it and I stopped doing CS for years until I had the opportunity to work on an app that I actually used every day. I was like, “Yeah, I want to edit that.” I just want to add this little checkbox there. Finally, I'll learn programming for that and relearn programming to do useful things for people. Motivation is key. SY: Yeah. I think because I started doing programming when I was quite young, I knew it was definitely the classic video games, wanting to learn how to make video games and then by the time I actually got to university, then I was like, “Yeah, don't want go into the games industry.” So didn't end up doing that. But I still enjoy game jams and things like that. If you're not again. CASEY: That's another thing you might like, Mandy. It's a weekend game jam. MANDY: Hm. CASEY: I don't know how into gaming you are, but it's also fun, lower stakes. People are just partying. Not unlike Ruby for Good. They happen more often and I like how it feels at a game jam, a little better than a hackathon because you're building something fun and creative instead of using a company's API because they told you to. SY: [laughs] Yeah. MANDY: Yeah, I was honestly never exposed to video games as a child. They were a no-no in my household and that's one of the things that I always cursed my parents for is the fact that I am the worst gamer. [laughs] My daughter makes fun of me. I'll sit down and like try to – she's 12 and I'll try to do something. She'll be like, “Wow, this is hurting me to watch you, Mom,” [laughs] and I'm like – [overtalk] CASEY: Ouch. MANDY: No, she called me a try hard and I was like, “Yeah, I'm trying really hard to just go forward.” Like I'm trying really hard to just jump over this object, [chuckles] I was like, “If that makes me a try hard well, then yes, I'm trying very hard. Thank you.” SY: Yeah. My 6-year-old has now got to the point where he can beat me at Super Smash Brothers so I'm not feeling too good about that. [laughs] CASEY: Yeah. My 6-year-old nephew beat us all in Mario Kart a couple weeks. SY: Yeah. [laughs] I can still beat in the Mario Kart. That, I could do. [laughs] MANDY: Yeah. A lot of the games she does looks fun, though so it's something I would be interested in, it's just something that I haven't been exposed to. I'm really excited now that—I don't want to say the pandemic is nearing an end because it seems to be not happening, but I'm excited – [overtalk] CASEY: True. Things are opening up. MANDY: Right now. Until they start closing down again. CASEY: Yeah. MANDY: Because I'm so excited for things like Ruby for Good, driving down to D.C. and seeing some of my friends, and I would be interested in going to one of those game things, as long as people are just like, “Oh yeah, we can be patient with her because she's never done a game before.” [laughs] CASEY: Yeah. My last game jam had eight people on the team and zero had ever done game development before. We figured something out. SY: [chuckles] Yeah. MANDY: Oh, that's fun. SY: Like muddle along. CASEY: Yeah. Somebody did like level design. They did a title map. Someone did sprites. They were like, “I'm going to do a sprite tutorial now.” Sprite is moving like a walking character. We had learned all the terms for it. We didn't know the terms either, but it was a good environment to learn. MANDY: It seems it. It seems like if you have a happy, healthy environment. For me, it was just, I was becoming stressed out. I had a standing meeting once a week with a really, really awesome person and it felt like it was more of like, I was like, “Oh my gosh, I have to work this into my already busy workweek and if I don't, then I'm completely wasting their time,” and I started to feel guilty to the point it brought me down. I was just like, “I don't think this is good for either one of us right now” because I'm feeling too much pressure, especially with the once-a-week thing and it's like to get through this chapter and then get through this chapter, and then I'd have a question and I'm not good at writing things down and then I'd forget. It seems like that might be more of a strategy to learn for me. I think a lot of people, there's different strategies like you have your visual learners, or you have your audio learners and I think for me, it would be cool just like I said, shadowing somebody. Like, if I just like sat there and it wasn't weird for me just to watch it over somebody's shoulder while they're doing this thing, that would a more conducive environment to the way I learn. CASEY: Yeah. I like the pattern, You do, We do, I do. Have you heard of that one? MANDY: No. CASEY: Or I do, We do, You do depending on the perspective. So it's like shadowing first and then doing it together where you're both involved and then you can do it on your own. It's a three-step process to make it a little bit easier to learn things from other people. SY: Yeah, that makes sense. MANDY: Yeah, that sounds like how kids learn. It's how we teach our children like I do, now we're going to do it together, now you do it. Yeah, I definitely have used that with my kid. [chuckles] CASEY: And it's just completely reasonable to do that as adults. That's how human brains work. MANDY: Yeah. No, I don't feel – that's the thing I would have to not almost get over, but just be like, “Oh my gosh, I'm 2 years old. I'm learning like I'm a toddler and that's so embarrassing.” But I think that that is a great way to learn and a great way to approach learning in general. I just started a book on learning more about crystals and it's the beginner's guide and she said, “You read this book and then you can move on to reading this other 700-page book that I've authored, but you should probably read this concise guide first.” I think a lot of people feel the pressure to dive into the super smart, or what they perceive as being the super smart way of diving in like, picking up the Ruby book, or the books that everyone talks about when there's so many other great resources exist that break it into smaller, bite-sized, digestible chunks. I think there's no shame in learning like that and I think a lot of people think that they just need to dive right in and be like, “Oh, this is the hard book, I'm going to go for the hard book first.” Like no, start with the easiest, start small. SY: Yeah. I think as you say, it definitely depends on how you learn what kind of resources you find interesting and engaging. CASEY: I've heard a similar story from a lot of friends, Mandy, where they really want to learn something, maybe programming in general, or a language, and then they psych themselves out, or they don't have the bandwidth in the first place, but they don't realize it and they struggle through that and the guilt because they want to, but they don't have time, or energy, which you also need. It's really common. A lot of people that I know are really motivated to do a lot of stuff; they want to do everything. I know some people who are fine not doing everything and that's great because they're probably more grounded. [chuckles] [laughter] But a lot of people I know really want to learn at all and it's a tension; you don't have infinite time and energy. SY: Yeah. I definitely fall into wanting to learn absolutely everything and right now. MANDY: So what kind of things are you teaching right now, Sy? What kind of content are you putting out there? SY: Yeah. So like I said, a lot of it's to do with low-level programming, like how memory actually works on a computer and how it affects how we program things. Because for a lot of people, if you come from a higher-level programming background, you're used to memory being abstracted away from what you do. You deal with variables, you deal with objects, and the implementation of the programming language deals with how that actually maps onto the underlying hardware. But if you really need to get the most performance you possibly can out of your system and you're using a little bit lower-level language like C, or C++, or Rust, or Swift, or something, then you need to understand how your processor is actually handling the instructions and that is actually handling your memory accesses in order for your performance to actually be good. Some of it is not obvious as well and does not match with how you might think memory works because the processors which we're using today are based in so much history and legacy. A lot of the time, they're essentially trying to mimic behavior of older processors in order to give us a programming model, which we can understand and work with, but then that means that they have to work in certain ways in order to actually get performance for the high-performance modern systems we need. So having an understanding of how our caches work, how instruction pipelines work, and things like that can actually make a really big difference down with the low-level programming. MANDY: Okay. So I'm looking at your Twitter and then looking at your pinned tweet, it says, “I made a YouTube channel for my ‘Computer Science Explained with my Cats' videos.” How do you explain computer science with your cats? Because that's something I could probably get into. SY: Yeah. So I have three cats and – [overtalk] MANDY: I've got you beat by one. SY: Nice. What were your cats called? MANDY: I have four. I have Nicks after Stevie Nicks. I have Sphinx because he looks so regal and I have Chessy and I have Jolie. SY: Cool. Mine are Milkshake, Marshmallow, and Lexical Analysis cat. MANDY: [laughs] Cool. SY: [chuckles] Yeah. So the things explained with my cats, it's mostly I wanted to explain things with my cats and random things, which I find around my house. So I remember I have a Discord server, which I help to moderate called #include , which is a welcoming inclusive organization for the C++ community. We were talking about hash maps and how hash maps are actually implemented, and I realized that there's a lot of different design areas in hash maps, which can be difficult to understand. I wanted to try and explain it using boxes and teddies and my cats so I set up a bunch of boxes. These are all of the buckets, which your items could go into it and then there's some way to map a given teddy to a given box. Let's say, it could be how cute it is. So if it's super cute and it goes in the west most box, and if it's kind of cute, then it goes into the box after that and so on and so forth. That's kind of how hash maps work. They have a bunch of memory, which is allocated somewhere, a bunch of boxes, and they have some way of mapping given items to a given box, which is called a hash function. In this case, it was how cute they are and then you have some way of what happens if two teddies happened to be as cute as each other, how do you deal with that? There's a bunch of different ways that you could handle that and that's called hash collision. Like, what do you do with collisions? Do you stick them in the same box and have a way of dealing with that, or do you just put them in the next box up, or a few boxes up, or something like that? There's whole decades worth of research and designing, which go into these things, but the concepts map quite nicely onto boxes and teddies and how cute they are. [chuckles] MANDY: I love that. SY: They are also explaining how caching works with chocolate, like the intuition with memory access is you ask for some chunk of memory and you get that chunks. You ask for a single chunk of chocolate and you get that chunk of chocolate, but in reality, that's not what happens in most cases. In most cases, you're actually going to get back a whole row of chocolate because it's most likely that if you're going to get a bit of chocolate, you're probably going to be accessing the bits which are right next to it. Like, if you have an array and you're processing all of the elements in that array, then you're just going to be stepping along all of those elements. So it's much faster to bring all of those elements would be right into memory at once. That's what happens in modern processors. Without you having to ask for it, they just bring in that whole row of chocolate. So I tried to – [overtalk] CASEY: That's so polite. [laughs] When your friend asks for a single chip, or a single piece of chocolate, you know what they want more. SY: [laughs] Yeah. CASEY: How generous of you to give them the whole bag. [laughs] Whether they want it, or not though. SY: Yeah. MANDY: So are these videos relatively short, or are they more long-form videos? SY: Yeah, they are 2 minutes long. MANDY: Oh, cool. SY: I try and keep them within the video limit for Twitter videos, which is 2 minutes, 20 seconds. MANDY: Okay, cool. See, that's something I could probably commit to is watching one of those videos not even maybe once a day because sometimes that's a little bit, much pressure every day. So maybe I try to work out three to four times a week. So saying I'm going to do this three to four times a week and I'm going to not stress on I'm going to do this every Monday. Generally three to four times a week, I think that's something I could, could commit to. SY: Yeah. Trying to get them within 2 minutes, 20 seconds can be really tough sometimes. Like it's quite – [overtalk] MANDY: Do you do a lot of editing? SY: Yeah. I would sit down and I'll write the whole episode, or video, or whatever and just get in all of the content that I want, just put it onto a text document and then I'll start filming it in whatever order I want, and then I start editing and then quite often, I realized that I've got 2 minutes, 40 seconds worth of content, or something and I can't quite cut it down and I have to reshoot something and then reedit it. I try to get it all done within a single day because if I don't get it done in a single day, then it ends up taking even longer because I get distracted and things like that. I need to focus just getting this one thing done. MANDY: So you're doing these within hours? SY: Yeah. MANDY: From start to finish, how many hours would you say you invest in these videos? SY: Start to finish, about 5, 6 hours, something like that. Like I said, I don't really have to do a lot of research for them because they're things I know very well, so I can pretty much sit down and just write something and then most of the time is spent in editing and then captioning as well. MANDY: Very cool. CASEY: I've been doing a bit of video editing lately and it takes so long. SY: Yeah, it really does. CASEY: I'm not surprised it takes 5, or 6 hours. [laughter] MANDY: No, I'm not either. I do all the podcasts editing. For those of you listening, who do not know, I edit all these podcasts and it takes roughly even 5 to 6 hours for audio, because I also put other work into that, like doing the show notes and getting the transcripts. Now I have those outsourced because I don't have enough hours in the day, but there's a lot of different parts to editing, podcasting, screen casting, and stuff that I don't think a lot of people know that these 2-minute videos that you do really do take 5 to 6 hours and you're putting these out there for free? SY: Yeah. MANDY: Wow. That's amazing. I assume you have a full-time job on top of that. SY: Yeah. Because my position is a developer advocate, I can count that as is doing work so I don't have to do that in my own time. MANDY: Very cool. Yeah, that's cool. I love DevRel so working in DevRel, I do that, too. I'm a Renaissance woman, basically. Podcast editing, DevRel conference organizing, it's a lot. SY: Yeah. MANDY: So I give you mad props for putting stuff out there and just giving a shout out to people who might not be aware that content creation is not easy and it does take time. So thank you. Thank you for that. Because this seems like the kind of stuff I would be able to ingest. SY: Yeah, thanks. MANDY: And that's cool. CASEY: I'm especially impressed, Sy that you have these interests that are complex would expand and you can explain the well and you find the overlap with what people want to know about. [chuckle] I think maybe in part from the Discord, you hear people asking questions. Can you tell us a little bit about what that's like? How do you decide what's interesting? SY: Yeah. I ask people on Twitter what they would find it interesting, but I also, because right now I'm not really going to conferences, but previously I'd go to a lot of conferences and people would come up to me and if I give a talk on compilers, for example, come and say like, “Oh hey, I never knew how register allocation worked. It was super interesting to know.” So I don't think I've done a video on register allocation yet actually. I should do one of those. MANDY: Write that down. SY: [laughs] Yeah. That's the kind of thing. Just because I spent a lot of time in communities, conferences, Discords, on Twitter, you get a feel for the kind of topics which people find interesting and maybe want to know how they work under the covers and just haven't found a good topic. Even function calls like, how does a function call work in C at the hardware level? If you call a function, what's actually happening? I did a video on that because it feels like such a fundamental thing, calling a function, but there's a lot of magic which goes into it, or it can seem like a lot of magic. It's actually, I want to say very well-defined, sometimes less so, but [laughs] they are real so there is random reason. MANDY: Very cool. I want to talk about the other content creation that you do. So code art journal and trashheap zine, do you want to talk about those a minute? SY: Sure. So code art was an idea that I had. It's a journal of code as art. I'd hear a lot of people saying, “Oh, coding is an art form.” I'd be like, “Okay. Yes. Sometimes, maybe. When is it an art form? When is it not? What's the difference between these?” Like, I spent a lot of time thinking about art because I'm a poet and I spend most of my free time researching and watching movies. Code as art is something which really interested me so I made this journal, which is a collection of things which people send in of code which they think is art and sometimes, it's something you might immediately see and look at it and think, “Okay, right, this is code and it's fulfilling some functional purpose,” and maybe that functional purpose gives it some artistic qualities just by how it achieved something, or if it does something in a very performant manner, or a very interesting manner. Other times, you might look at it and say, “Okay, well, this is code, but it's more aesthetic than functional.” And sometimes it's things which you might look at and think, “Okay, is this even code?” Like there was someone sent in a program written in a language called Folders, which is a esoteric programming language entirely programmed using empty folders on your hard drive, which I absolutely love. I'm super into esoteric programming languages so I absolutely loved that one. [chuckles] But yeah, so the – [overtalk] CASEY: That sounds so cool. Where can people find it? Is it online also? SY: Yes, it's in print and there's also, you can get the issues online for free in PDF form. There is a third issue, which is pretty much fully put together on my machine, I just haven't done the finishing touches and it's been one of those things that's just sat, not doing anything for months and I need to get finished. [chuckles] And then trashheap zine is another thing that I co-edit, which is just utter trash, because as much as I love more explicitly artistic films and writing and things like that, I also have a deep love of utter, utter trash. So this is the trashiest stuff that we could possibly find, even the submission guidelines that I wrote for that is essentially a trash pond, but random submission guidelines. So if you have trash, please send our way. MANDY: Yeah. I was going to say, what you consider trash? What trashiest [laughs] enough to be in these zines? SY: I can read out, where's my submission guidelines? The URL for the zine is trashyheap.party, which I was very, very pleased with and the website looks awful. I spent a lot of time making it as awful as I possibly could. Things like any kind of – [overtalk] CASEY: I love the sparkles. SY: Yes! CASEY: When the mouse moves, it sparkles. SY: Isn't it the best, seriously? Yeah. CASEY: Every website should have that. SY: Yeah, totally. Like texts you sent your crush at 4:00 AM while drunk where you misspelled their name and they never spoke to you again, or draft tweets which you thought better of sending, purely Photoshop pictures of our website. [laughter] A medically inaccurate explanation of the digestive system of raccoon dogs. All good stuff. MANDY: That's amazing. CASEY: I know a lot of people who would be cracking up reading this together. [laughter] CASEY: That sounds great. There's so much treasure in this trash heap. MANDY: Yeah. Don't worry, folks, we'll put links in the show notes. CASEY: Oh, yeah. SY: Yeah. One of my favorite things with it was when we'd get all of the submissions, we would get together and just project them up on a wall and read them together and so much so bad, it's hilarious in the most wonderful way. CASEY: That sounds like a party itself. SY: It is, yes. CASEY: The be trashheap party. SY: Absolutely. CASEY: It's kind of taking me back to early pre-YouTube internet when we watch flash cartoons all the time and a lot of those were terrible, but we loved them. SY: Yes. I made some as well, they were so bad. [laughter] I remember getting a very non legal version of flash and making the worst stick flash renovations I possibly could. CASEY: Oh, speaking of content creation, I've been learning some animation and 3D modeling animation lately. I had my first ever viral TikTok; it had over 9,000 views. SY: Wow! Nice. CASEY: And so when I look at my phone, if it's not the notifications muted, it's annoying. I have to turn it off. [laughter] SY: Yeah – [overtalk] MANDY: Congratulations! [laughs] CASEY: Thank you. So the video is a USB thumb drive that won't insert, even though you flip it over. That's been done before, but what I added was misheard lyrics by the band Maroon 5. Sugar! USB! That's what I hear every time. Mandy, have you done any art? MANDY: Have I done any art? CASEY: Lately? MANDY: Oh. Yeah. Well, actually – [overtalk] CASEY: You've been doing some home stuff, I know. MANDY: Yeah. I've been doing plant stuff, gardening, but this weekend, I actually took my daughter to a workshop. It was called working with resin—epoxy. SY: Oh, cool. MANDY: And we got to make coasters. The teacher brought stickers, feathers, and crystals and it was like a 3-hour workshop and I think my daughter had extra resin. Her birthday is on Thursday this week and I noticed she was making kind of the same ones and I said, “What are you doing?” And she said, “I'm making gifts for my friends that come to my birthday party.” I just thought it was so sweet that I was like – [overtalk] SY: Oh, so sweet. MANDY: Usually birthday parties, you receive gifts, or whatever and she's like, “No, I would like to give them gifts for my birthday,” and I was like, “Oh, that's adorable.” So I've been trying to do more things with my hands and get off the screens more, which has been the major thing keeping me back from being on code. I've made a strict weekend policy where I do not touch my computer from Friday evening to Monday morning, unless it's an absolute dumpster fire, I need to do something, or if a takeout menu looks better on my computer than it does on my phone. [laughter] Then I'll pop it open, but I won't read the email, or do the Slack. And then this Saturday I'm taking a course in astrology. It's all-day workshop so I'm excited to kind of dive into that stuff a little bit more. CASEY: So cool. It's hard to believe we can do these in person again. I'm not over it. MANDY: I know. I'm so afraid to get excited over it and then have it be taken away again. CASEY: Yeah. Sy, tell us a little more about #includes . I've actually heard of it. It's a little bit famous online. It's an inclusive community, I know from the name. SY: Yes. CASEY: Tell us more about it. SY: So it actually started off on Twitter as a half joke; Guy Davidson tweeted being like, “Hey, so why isn't there a diversity and inclusion organization for C++ called #include?” Because #include is it's like a language concept in C and C++ and people were like, “Hahaha yeah, you're right,” and then Kate Gregory was like, “You're right. We should make one.” So we did [chuckles] and we started off with like six of us in a Slack channel and then ended up moving to Discord and starting our own server there and now we are a few thousand members. Back when we had in-person conferences, we would have a booth at pretty much every major C++ conference, we had scholarships, which we would send people on, we got conferences to improve by having live captioning and wheelchair accessible stages and gender-neutral bathrooms instituting and upholding code of conduct, things like that. We started off thinking, “Hey, if we could get some conferences to have a code of conduct or something that would be great,” and then it ended up being way, way, way bigger than any of us thought it would become, which is amazing to see. CASEY: That's so cool. What a success story. SY: Yeah. CASEY: How long has it been going on now? SY: I guess about 3, or 4 years. Yeah, probably closer to 4 years. My sense of time is not good the best of times, but something around 4 years. CASEY: I'm curious if another language community wanted to do something similar if they're inspired. Is there a writeup about what y'all have done? 
SY: I've given talks. CASEY: That we can point people to. We can put that in the show notes. SY: Yeah. I've given a couple of talks, as I said. CASEY: Talks, that would be good. SY: Other people have given talks as well. I gave a slightly longer form talk DevRelCon, London in 2019, I think, which was on the lessons which we learned through trying to build a welcoming and inclusive community. Community which has already been around for decades because C++ was first standardized in 1998 so it's been around for quite a long time and has a lot of history. CASEY: That sounds great. I can't wait to watch it. SY: Yeah. I know that there's other languages. You have JavaScript, QueerJS, which is a really cool community and I'm sure there are other languages which have similar things going as well. CASEY: I had never heard of QueerJS. I'm queer and JS. SY: Yeah. CASEY: I'm glad I had this moment just now. SY: It's cool. They have a Discord and I can't remember how active the Discord is, but they would have meetups across the world, they have one in London and in Berlin and bunch of other places, and talks and community. It seems really cool. CASEY: That's awesome. SY: I wanted to give a talk about C++ and JavaScript because you could link target JavaScript with C++ these days, which is kind of cool. CASEY: I've used Emscripten before. SY: Yeah. CASEY: I didn't use it directly, other people did. It turned Graphviz into a JavaScript. A program that runs in JavaScript instead of normally, it's just CSS. So I could draw circles pointing to other circles in the browser, which is what I always wanted to do. Graphviz.it, that “it” is my favorite Graphviz editor. It's online. SY: Cool. I like Graphviz a lot. Emscripten is really cool, though. Basically a way of compiling C++ plus to JavaScript and then having the interoperation with the browser and the ecosystem that you might want to be able to call JS functions from C++, or other way around, and do things which seem operating systems E, but have to be mapped inside the browser environment. CASEY: That's powerful. I'm also glad I've never had to use it directly. Other people made libraries doing it what I needed. Thank goodness. [chuckles] Abstraction! SY: Yeah. I've not used a whole lot, but I did find it fairly nice to work with when I did. I made a silly esoteric programming language called Enjamb, which is a language where the programs are cones and it runs on a stack-based abstract machine and the interpreter for it is written in C++. I wrote a command line driver for it and also, a version which runs in the browser and that compiles using Emscripten. It was really cool and I picked it all up with CMake, which is the main C++ build systems that you could just say, “Hey, I want to build the combine line version for my platform” like Windows, or Mac, or Linux, or whatever, or “Hey, I want to build it for the web,” and it would build the JavaScript version in HTML page and things like that. It's pretty cool. I recently made another esoteric programming language, which you program using MS Paint. You literally make shapes with MS Paint and you give the compiler an image file, and then it uses OCR and computer vision in order to parse your code and then generate C from that. [laughs] It's pretty ridiculous, but I had so much fun with it. CASEY: OCR is Optical Character Recognition? SY: Yes, exactly. CASEY: So I'm picturing if I wrote a program on a napkin and a computer could maybe OCR that into software. SY: Yeah. So it uses OCR for things like function names because it supports function calls and then uses shapes for most things. It has things like a plus sign, which means increment what it's currently being pointed to, or right, or left, or up, or down arrow is for moving things around. You would actually make an image file with those symbols and then I used OpenCV for working out what the shapes were. It was the first time I've ever done any kind of image recognition stuff. It was a lot easier than I expected it to be; I thought we'd have to write a lot of code in order to get things up and running and to do image detection. But most of the simple things like recognizing hey, this is a triangle, or this is a plus sign, or this is a square, and things like that were pretty, you don't need a lot of code in order to do them. That was mostly when you had to say like, “Okay, this is a triangle, but which direction is it pointing in?” It got a little bit more complicated; I had to do some maths and things like that and I'm terrible at maths. [chuckles] So that was a little bit more difficult, but it was a lot fun to get started with and I had a much lower barrier to entry than I expected. CASEY: Now I want to play with OCR and image recognition. I haven't done that for 10 years. It was not easy when I tried it last time with whatever tool that was. SY: [chuckles] Yeah, I did it – [overtalk] CASEY: For the future! SY: [laughs] Definitely. Yeah. I did it with Python and Python has fairly nice OpenCV bindings and there's a ton of resources out there for predicting most of the basic stuff that you would expect. So there's a lot of learning resources and decent library solutions out there now. CASEY: Cool. All right. We're getting near the end of time. At the end, we like to go through reflections, which is what's something interesting that stood out to you, something you'll take with you going forward from our conversations today. MANDY: I really am excited to dig into Sy's videos. They seem, like I said earlier in the show, something I could commit to a few times a week to watching these videos especially when they are concepts that seem so much fun, like cats, teddy bears, cuteness levels, and things like that. I think that would be a great start for me just to in the morning while I'm still drinking tea just before I even dive into my email, check out one of those videos. So I think I'll do that. SY: Thanks. CASEY: Sy, I liked hearing about your process side with your constraints like 2 minutes, 20 seconds on Twitter, that's such a helpful constraint to make sure it's really polished and dense. It takes you 5 to 6 hours and you make things that people ask about, that they're interested in. That whole process is fascinating to me as I try to make more viral TikToks. [laughter] Or whatever I'm making at the time. SY: Yeah. CASEY: I always wondered how you made such good stuff that got retweeted so often. Cool things of insight. SY: Yeah. Mostly just time. [laughs] I guess, it makes me remember that I definitely want to make a video on register allocation because I love register allocation. It's such a cool thing. For those who don't know, it's like if you have a compiler which takes your code and maps it onto the hardware, your hardware only has a certain number of resources so how do you work out how to use those resources in the best manner? It maps onto some quite nice computer science algorithms like graph coloring, which means it maps quite nicely visually, I could probably make a pretty cool graph coloring visualization with some random things I have strewn around my room. CASEY: I can't imagine this yet, but I will understand that clearly soon I bet. MANDY: That's awesome. Well, I just want to wrap up by saying thank you so much for joining us today, Sy. This has been a really awesome conversation. And to folks who have been listening, thank a content creator. It takes time. It takes energy. It's a lot of work that I don't think a lot of people, unless you've done it, really understand how long and in-depth of a process it is. So thank one of us content creators, especially when we're putting this content out for you for free. To do that for us Greater Than Code, we do a Patreon page and we will invite Sy to join us and we would like you to join us as well. If you are able to donate on a monthly basis, it's awesome. It's patreon.com/greaterthancode. All episodes have show notes and transcripts, and we do a lot of audio editing. So join us if you're able. If you are still a person who is greater than code and cannot afford a monthly commitment, you are still welcome to join us in our Slack community. Simply send a DM to one of the panelists and we will let you in for free. So with that, thank you so much, Casey. Thank you again, Sy. And we'll see you all next week. Special Guest: Sy Brand.

This week's EYE ON NPI is moving in the right direction - it's U-Blox's XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit! This NPI tries to solve a well-trod but still not-fully-solved technical challenge of how to perform 3D tracking of low cost/low power 'tags' indoors. Ironically, object tracking is something that humans and animals do very well - we can even track things that have hidden themselves from view! But for robotics, this is an incredibly hard problem. Let's talk about some of the ways that we can do object tracking now, to explain why something that sounds so simple has been a challenge for decades. The way robotics do object tracking now is how the most simplistic organic vision systems work: by looking for a contrasting color or shape. This works best if something is vividly colored - like an Aibo robotic dog tracking a round pink ball (https://commons.wikimedia.org/wiki/File:AIBO_ERS-7_following_pink_ball_held_by_child.jpg#filelinks). Works great if the thing you're trying to track happens to be round and pink, which is not many things on this planet. Also, doesn't work particularly well if the shape is hidden or obscured. The next step up from basic shape tracking is machine learning vision systems that try to recognize objects by using layers of matrix calculations - what we would normally call OpenCV (https://en.wikipedia.org/wiki/OpenCV) or TensorFlow (https://en.wikipedia.org/wiki/Object_detection#/media/File:Detected-with-YOLO--Schreibtisch-mit-Objekten.jpg). This is a little more flexible, but still requires good lighting, unobscured vision, and recognizable shapes. If you don't want to use a camera, and you happen to be outside, you can use GNSS/GPS. A board with a GPS module and a radio transceiver can pretty easily determine location and then relay it back to a central station. GNSS gives you up to 10 meters This is great for cars, people, boats - all sorts of large objects. But that 10 meter precisions makes it tough for smaller items and of course, GNSS does not work indoors. (If you do want better precision, you can get it using RTK - check out our EYE ON NPI from last year https://blog.adafruit.com/2020/06/08/eye-on-npi-u-blox-c099-f9p-application-board-for-zed-f9p-gnss-rtk-module-eyeonnpi-adafruit-digikey-digikey-ublox/) Traditionally, when working indoors to do tracking, folks have relied on a few different technologies. In particular RSSI tracking is quite popular because its so cheap. Basically, radio signal strength falls geometrically with distance from antenna to antenna. Folks can also use time-of-flight technology, which has slowly been making it to WiFi modules (and may also make it into BTLE at some point - we'll do EYE ON NPI on that technology when it filters into the market!) As of Bluetooth 5.1, there's a new Direction Finding capability built into the wireless specification. Bluetooth direction finding makes it possible to determine the direction that radio signals travel from a mobile tag to one or several fixed anchor points. Using angle-of-arrival (AoA) technology, anchor points comprising an antenna array that is connected to a Bluetooth receiver can detect the direction, or angle, to the mobile tag, which transmits a Bluetooth signal. When a constellation of such multi-antenna anchors is deployed, AoA technology can be used to triangulate the precise location of a mobile device or tag. (https://www.u-blox.com/en/press-releases/u-blox-presents-bluetooth-aoa-explorer-kits-high-precision-indoor-positioning) Note that this is not distance measurements, it's angular measurements. But, of course - if you have a few fixed antenna station locations its easy to convert a set of angles into a precise location! The angle calculations seem to give better accuracy than plain RSSI - 1 to 2 meters - and can be used for more than just location sensing. For example, in this demo from U-Blox, a camera can follow a tag just with angular data since we don't care how far away the target is, just that it is in frame (https://www.u-blox.com/en/blogs/tech/how-we-built-our-bluetooth-direction-finding-demo) Direction Finding is included in any U-Blox/Nordic module that supports BLE 5.1, but to make it easy we recommend picking up a XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit (https://www.digikey.com/short/q3dpn32p) that has the antenna configuration needed and laid out. It's a lot easier and faster than routing your own boards - the 'tags' are any BLE module and do not need special design considerations. Lucky for us, the U-Blox XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit (https://www.digikey.com/short/q3dpn32p) is in stock at Digi-Key right now, and is an excellent way to get started immediately with trying out the new technology. If you want a more advanced setup, with 4 fixed-point-nodes and 4 tags, sign up for the XLPR-AOA-2 kit (https://www.digikey.com/en/products/detail/u-blox/XPLR-AOA-2/14666759).

Hackaday editors Elliot Williams and Mike Szczys celebrate the cleverest projects from the week that was. We tried to catch a few fools on Thursday with our Lightmode™ and NFT articles -- make sure you go back and read those for a good chuckle if you haven't already. While those fall under not a hack, many other features this week are world-class hacks, such as the 555 timer built from 1.5-dozen vacuum tubes, and the mechanical word-clock that's 64 magnetic actuators built around PCB coils by Hackaday's own [Mortiz v. Sivers]. A treat for the ears, [Linus Akesson] aka [lft] shows off a Commodore64 that seriously sounds a good as a cathedral organ. And a masterpiece of OpenCV and Blender, you can't miss the project by [Matthew Earl] that overlays video of the Mars landing on still satellite photos... perfection! Check out the show notes!  

In the Blind Abilities Studio, we welcome Jagadish K. Mahendran, Artificial Intelligence Developer and Engineer,and Hema Chamraj, director, Technology Advocacy and AI4Good at Intel Raqi joins Jeff in the studio to learn and find out more about this great initiative that may one day enhance the  navigating experience while Blind. From the Intel Press Release: Intel just announced a research project involving an AI-powered backpack that can help the visually impaired navigate and perceive the world around them with voice commands. Artificial intelligence (AI) developer Jagadish K. Mahendran and his team designed an AI-powered, voice-activated backpack that can help the visually impaired navigate and perceive the world around them. The backpack helps detect common challenges such as traffic signs, hanging obstacles, crosswalks, moving objects and changing elevations, all while running on a low-power, interactive device. “Last year when I met up with a visually impaired friend, I was struck by the irony that while I have been teaching robots to see, there are many people who cannot see and need help. This motivated me to build the visual assistance system with OpenCV’s Artificial Intelligence Kit with Depth (OAK-D), powered by Intel.” – Jagadish K. Mahendran, Artificial Intelligence Engineer The World Health Organization estimates that globally, 285 million people are visually impaired. Meanwhile, visual assistance systems for navigation are fairly limited and range from Global Positioning System-based, voice-assisted smartphone apps to camera-enabled smart walking stick solutions. These systems lack the depth perception necessary to facilitate independent navigation. “It’s incredible to see a developer take Intel’s AI technology for the edge and quickly build a solution to make their friend’s life easier,” said Hema Chamraj, director,  Technology Advocacy and AI4Good at Intel. “The technology exists; we are only limited by the imagination of the developer community.” The system is housed inside a small backpack containing a host computing unit, such as a laptop. A vest jacket conceals a camera, and a fanny pack is used to hold a pocket-size battery pack capable of providing approximately eight hours of use. A Luxonis OAK-D spatial AI camera can be affixed to either the vest or fanny pack, then connected to the computing unit in the backpack. Three tiny holes in the vest provide viewports for the OAK-D, which is attached to the inside of the vest. The OAK-D unit is a versatile and powerful AI device that runs on Intel Movidius VPU and the Intel® Distribution of OpenVINO™ toolkit for on-chip edge AI inferencing. It is capable of running advanced neural networks while providing accelerated computer vision functions and a real-time depth map from its stereo pair, as well as color information from a single 4k camera. A Bluetooth-enabled earphone lets the user interact with the system via voice queries and commands, and the system responds with verbal information. As the user moves through their environment, the system audibly conveys information about common obstacles including signs, tree branches and pedestrians. It also warns of upcoming crosswalks, curbs, staircases and entryways. More Context: A Vision System for the Visually Impaired (Case Study) | Intel OpenVINO Toolkit | Artificial Intelligence at Intel | MIRA Contact Your State Services If you reside in Minnesota, and you would like to know more about Transition Services from State Services contact Transition Coordinator Sheila Koenig by email or contact her via phone at 651-539-2361. Contact: You can follow us on Twitter @BlindAbilities On the web at www.BlindAbilities.com Send us an email Get the Free Blind Abilities App on the App Storeand Google Play Store. Check out the Blind Abilities Communityon Facebook, the Blind Abilities Page, and the Career Resources for the Blind and Visually Impaired group

In this article, I will show you how to do noise reduction in 3 simple steps. We will be using a machine learning trained noise reduction model. It is one of the best noise reduction models I've found out there. I will share more about the model and how to apply it in the following paragraphs. In this project, we will be using three python packages. The packages are as follows: OpenCV, Matplotlib, and NumPy. OpenCV is a very well-known kit for computer vision. As a prerequisite for OpenCV library, we will need to install Numpy. We transform pixels into arrays when reading an image; NumPy is going to do that in the behind scenes. When dealing with multi-dimensional arrays, NumPy is perfect. This episode is also available as a blog post: http://sonsuzdesign.blog/2021/02/27/enhance-your-images-using-opencv-noise-reduction-algorithm/

En este capítulo hablo de cómo instalar OpenCV en una Raspberry Pi para poder crear programas y algoritmos de Visión Artificial.Puedes ver el tutorial completo en https://programarfacil.com/blog/vision-artificial/opencv-raspberry-pi/

Developer, Instructor, Teacher, Author, Conference Speaker, Google Developer Expert, YouTuber, Podcaster and man with a voice totally suited for crime solving serial podcasts… Ladies and gentlemen, today’s guest is Nate Eble. Oh btw, in the podcast I say OpenCV is Open Computer View. It's actually Open Source Computer Vision. Apologies for the mix-up folks! It was an off day for me. Goobar - goobar.io Goobar on YouTube - youtube.com/goobar Goobar Podcast - goobar.buzzsprout.com OpenCV - opencv.org 360AnDev Conference: 360andev.com Mastering Kotlin - amzn.to/35nF0vu Kotlin First – Taking Full Advantage of Kotlin For Android Development - 360andev.com/session-recordings/?vimeography_gallery=3&vimeography_video=444346415

This weeks EYE ON NPI gets REAL with the Intel RealSense LIDAR camera. (https://www.digikey.com/en/product-highlight/i/intel/lidar-depth-camera-l515) This surprisingly powerful camera is easy to use and has both color and LIDAR depth output data. This makes it perfect for use in robotics, interactive art, 3D scanning...all sorts of projects where spatial sensing is required. It was announced in 2019 but didn't start shipping until a few months ago. Almost 10 years ago, Adafruit sponsored a Kinect-hacking bounty (https://blog.adafruit.com/2010/11/10/we-have-a-winner-open-kinect-drivers-released-winner-will-use-3k-for-more-hacking-plus-an-additional-2k-goes-to-the-eff/) to take an off-the-shelf video game accessory and turn it into a tool for anyone to use. The open source cross-platform drivers spawned hundreds of awesome projects, from the scientific and artistic, to the humanist and just plain silly (https://blog.adafruit.com/category/kinect-hacking/). With the Kinect discontinued (https://en.wikipedia.org/wiki/Kinect), it's been a challenge to find a low cost alternative with an easy API. Until now! This little sensor is smaller than a tuna can and is a complete system with plug-and-play USB type C port for power and data. There's even an BMI085 IMU inside! (https://www.digikey.com/product-detail/en/bosch-sensortec/BMI085/828-1083-1-ND/8634943) Intel RealSense's L515 is a type of solid state LiDAR technology-based depth camera. It uses LiDAR technology to scan a full scene allowing it to provide a point cloud with up to 23 million points of accurate depth data per second. It delivers consistent high depth accuracy throughout the supported range of the camera starting at 25 cm up to 9 meters. The accuracy of the L515 within its operational range is less than 20 mm Z error at maximum range. With an internal vision processor, L515 achieves motion blur artifact reduction by an exposure time of less than 100 ns (nanoseconds), ideal for capturing rapidly moving objects with minimal motion blur. Additionally, the 4 ms photon to depth latency allows real-time applications such as autonomous navigation. An RGB camera, gyroscope, and accelerometer round out the capabilities of the device for high-quality performance and use case flexibility. Smaller than a tennis ball, the Intel RealSense LiDAR Camera L515 has a diameter of 61 mm and is 26 mm in height. Weighing in at around 100 g (3.5 oz), it is designed to be easily situated on any prototype or attached to a tablet or phone for use in handheld room scanning or volumetric measurement applications. The L515 uses the same open source Intel RealSense SDK 2.0 as all other current-generation Intel RealSense Depth Cameras, which is platform independent, supporting Windows, Linux®, Android™, and macOS. This includes wrappers for many common platforms, languages and engines, including Python, ROS, C/C++, C#, Unity, Unreal, OpenNI, and NodeJS, with more being added constantly. Features Depth FOV (field of view): (H × V) 70°±2° x 55°±2° Depth resolution: Up to 1024 x 768 depth resolution (XGA resolution) Depth frame rate: 30 frames per second RGB frame rate: 30 frames per second RGB resolution: Up to 1920 x 1080 RGB resolution RGB FOV: (H x V) 70° ±3° x 43° ±2° Use environment: Indoor Inertial measurement unit: Bosch BMI085 Range: From 25 cm to 9 m range Connection type: USB-C® 3.1 gen 1 connection Dimensions: 61 mm diameter x 26 mm height Mounting: Two M3 x 0.5 mounting points and one 1/4 20 UNC thread mounting point It shows up in a few ways when you plug it into your computer. For integration into projects that use OpenCV or other camera-based inputs, you will see a dual UVC camera one called 'Depth' which is the LIDAR and one called 'RGB' which is a regular camera that you can overlay over the depth data. The Cameras are really high res, as well - 1024x768 depth resolution and 1920 x 1080 RGB resolution. Compare that to the Kinect's 640x480! (https://en.wikipedia.org/wiki/Kinect) We like that this sensor has industrial-quality sensing with engineer-and-maker-friendly development. No NDAs for datasheets, no signups for code - everything needed to get set up was shockingly convenient. (https://www.intelrealsense.com/get-started/) There's even a developer center (https://www.intelrealsense.com/developers/) with firmware upgrades, documentation, code samples, and a community group. There's a lot of support, so we think this is a great option for low cost depth sensing - a really fabulous NPI! Interested? Of course, you are! Pick up an Intel RealSense camera from Digi-Key today, it's part number 2311-82638L515G1PRQ-ND, and you can have it ready for your next amazing project tomorrow morning! (https://www.digikey.com/product-detail/en/intel-realsense/82638L515G1PRQ/2311-82638L515G1PRQ-ND/11688015)

An ML webapp usually requires:a) Tensorflow b) Opencv c) Flask

Video Version: https://youtu.be/wyfROwJUW-Y Subscribe here to the newsletter: https://tinyletter.com/sanyambhutani In this episode, Sanyam Bhutani interviews the Computer Vision Guru, Chief at PyImageSearch: Dr. Adrian Rosebrock. This interview is part-2 of Sanyam's blog interview with Adrian. They talk about Adrian's journey into CV and ML. They also discuss the secrets of PyImageSearch HQ and how the amazing tutorials are created. They also talk about Adrian's upcoming OCR Book and the importance of OCR. Links: OCR Book (Indiegogo link): https://www.indiegogo.com/projects/ocr-with-opencv-tesseract-and-python/ PyImageSearch: https://www.pyimagesearch.com Courses and Books by Adrian: https://www.pyimagesearch.com/books-and-courses/ Previous Interview: https://hackernoon.com/interview-with-the-author-of-pyimagesearch-and-computer-vision-practitioner-dr-adrian-rosebrock-e00583a225a0 Follow: Adrian Rosebrock: Sanyam Bhutani: https://twitter.com/bhutanisanyam1 Blog: sanyambhutani.com About: https://sanyambhutani.com/tag/chaitimedatascience/ A show for Interviews with Practitioners, Kagglers & Researchers and all things Data Science hosted by Sanyam Bhutani. You can expect weekly episodes every available as Video, Podcast, and blogposts. If you'd like to support the podcast: https://www.patreon.com/chaitimedatascience Intro track: Flow by LiQWYD https://soundcloud.com/liqwyd #OpenCV #PyImageSearch #ComputerVision

In this episode #27, the hosts Naveen Samala & Sudhakar Nagandla have interacted with another guest Yaswanth. Yaswanth Sai Palaghat is currently working as a software engineer at a product based company in Hyderabad. A machine learning Evangelist with a major content delivery focus on machine learning. Owns a python library developed during his engineering.  Active GitHub user with more than 150 real-time projects available in GitHub. He considers himself as a Techie, loves exploring technologies and teaching the same in his community. He is an Udemy instructor with active courses. Yashwant is a YouTuber with a well-strengthened YouTube channel with the motto of delivering tech content and will always try to motivate engineering students with his videos. A content developer with active tech content delivery in all forms(text, graphics, videos).  A blogger and runs an active blog titled "TECHIE EMPIRE" which aims to help and motivate the engineering graduates to choose the right path on technologies. He is very active in social media with more than 20k connections all over his social profiles and pages in LinkedIn, Instagram and YouTube. Coming to his education, Yaswanth graduated from Narayana Engineering College in the stream of Computer Science and Engineering in 2019 in which he was a college topper.  Apart from tech life, he is interested in film making where he is into story writing, editing, direction and a protagonist too. He Owns a YouTube channel named "THE FUN BITE" with more than 10 short films. Listen to Yaswanth's guidance on: How to Explain Machine Learning to your grandparents? How ML can help in solving real world problems? How to start leaning ML? Resources for learning ML Is it important to have mathematical background for ML? ML vs Deep Learning – Basic differences How ML is related to AI & Data Science? Yaswanth's LinkedIn profile: https://www.linkedin.com/in/yaswanthpalaghat/ Resources shared by Yaswanth: 1. Linear Algebra & Statistics: Linear Algebra for ML :https://youtu.be/1VSZtNYMntM Statistics for ML : https://youtu.be/hjZJIVWHnPE 2. Python Programming:    Python for Absolute Beginners: https://www.udemy.com/course/pythonforabsolutebeginners/?referralCode=E2DBB8598710151C2783 3. Data Analytics with Python Libraries(Numpy, Pandas, Matplotlib):     Data Analytics A-Z with Python: https://www.udemy.com/course/dataanalyticsa-zwithpython/?referralCode=AB085EE2CA864CB69FD7 4. Data Cleaning:    https://youtu.be/2HzsTwHL7H0 5.Project-I     Data Analytics on Iris Flowers Dataset : https://github.com/yaswanthpalaghat/Pandas-implementation-and-workflow-on-iris-flowers-dataset   6.Computer Vision with OpenCV:    Understanding OpenCV : https://opencv.org/  7.Project-II    Building a Face Detection and Recognition Model : https://www.udemy.com/course/building-a-face-detection-and-recognition-model-from-scratch/?referralCode=28F5323F045A89665F90 8.Machine Learning with Scikit-Learn: https://youtu.be/pqNCD_5r0IU 9.Hands-on with ML Algorithms: https://youtu.be/RnFGwxJwx-0 10.Solving ML Challenges:   https://www.kaggle.com/learn/microchallenges 11.Project-III: Linear Regression on Diabetes DataSet :    https://github.com/yaswanthpalaghat/Linear-Regression-on-Diabetes-data-set-to-find-out-the-predicted-data-set 12. Natural Language Processing and Text analytics:  https://youtu.be/OQmDhwhj78Y 13.Project-IV      Sentiment Analysis using NLP : https://github.com/yaswanthpalaghat/NLP-Sentiment-analysis-using-Machine-Learning-and-flask Enjoy the episode! Do not forget to share your suggestions or feedback at theguidingvoice4u@gmail.com or by messaging at +91 9494 587 187  Subscribe to our YouTube Channel: https://www.youtube.com/c/TheGuidingVoice Also, follow The Guiding Voice on Social Media: LinkedIn: https://www.linkedin.com/company/theguidingvoice Facebook: http://facebook.com/theguidingvoice4u Twitter: http://twitter.com/guidingvoice Instagram: https://www.instagram.com/theguidingvoice4u/ Pinterest: https://in.pinterest.com/theguidingvoice4u/pins/  #growth #data #ML #Machinelearning #supervisedlearning #unsupervisedlearning #deeplearning  #Neuralnetwork #ANN #datascience #datascientist #machinelearning #AI #AIbasics #technologies #kdnuggets #career #jobs #careerguidance #mentorship #careerpath #progression #management #leadership #crisis #job #midcareer #youngprofessionals #careergraph #TGV #theguidingvoice #kaggle #analyticsvidya #udemy #insofe #coursera #opensource #transformation    

Дмитрий Куртаев - разработчик в компании Intel и Core Team Member проекта OpenCV Профиль Дмитрия в twitter: https://twitter.com/dkurtaev Шоуноты: - О российском офисе Intel в Нижнем Новгороде (02:25) - Разработка Parallel STL (06:30) - Работа над компьютерном зрением в Intel (07:44) - Развитие OpenCV и области применимости компьютерного зрения (10:55) - Безопасно ли использовать компьютерное зрение в области self-driving cars (20:18) - Как развивается С++ сегодня. Почему в OpenCV не используется самый последний стандарт C++ (23:18) - Является ли Rust заменой C++ (30:05) - C++ как лучший второй язык для разработчиков (34:44) - IT-сообщества в Нижнем Новгороде (41:13) - Плюсы и минусы жизни в Нижнем Новгороде (45:33) - Совет от гостя (49:58) Полезные ссылки: IT-сообщество Нижнего Новгорода - https://www.it52.info/ C++ as a second language (Chrome University 2019) - https://www.youtube.com/watch?v=cN9c_JyvL1A Parallel STL - https://software.intel.com/en-us/articles/get-started-with-parallel-stl OpenCV - https://opencv.org/ Летняя школа Intel - https://russia-students.ru/isi Telegram—канал CSSSR: https://t.me/csssr Twitter CSSSR: https://twitter.com/csssr_dev Telegram ведущего: https://t.me/sgolovin Twitter ведущего: https://twitter.com/_sgolovin Telegram редакции: https://t.me/Vindizh Twitter редакции: https://twitter.com/Vindizh

Show Notes:(2:12) Parul talked about her educational background, studying Electrical Engineering at the National Institute of Technology, Hamirpur.(3:18) Parul worked as a Business Analyst at Tata Power India for 7 years.(4:29) Parul talked about her initial interests in writing about data science and machine learning on Medium.(6:30) Parul discussed her first blog series “A Guide to Machine Learning in R for Beginners” - which covers the Fundamentals of ML, Intro to R, Distributions and EDA in R, Linear Regression, Logistic Regression, and Decision Trees.(8:02) Reference to her articles on data visualization, Parul talked about matplotlib, seaborn, and plotly as the main visualization libraries she practices, in addition to Tableau for building dashboard.(10:11) Parul shared her thoughts on the state of Machine Learning interpretability, in reference to her articles on this topic.(13:54) Parul discussed the advantages of using Jupyter Lab over Jupyter Notebook.(17:30) Parul discussed the common challenges of bringing recommendation systems from prototype into production (Read her two articles about recommendation systems: (1) an overview of different approaches and (2) an overview of the process of designing and building a recommendation system pipeline)(21:00) Parul went in depth into her NLP project called "Building a Simple Chatbot from Scratch in Python (using NLTK).”(23:26) Parul continued this chatbot project with a 2-part series on building a conversational chatbot with Rasa stack and Python and deploying it on Slack.(28:15) Parul went over her Satellite Imagery Analysis with Python piece, which examines the vegetation cover of a region with the help of satellite data.(32:22) Parul talked about the process of Recreating Gapminder in Tableau: A Humble Tribute to Hans Rosling.(35:17) Parul discussed her project Music Genre Classification, which shows how to analyze an audio/music signal in Python.(39:20) Parul went over her tutorials on Computer Vision: (1) Face Detection with Python using OpenCV and (2) Image Segmentation with Python’s scikit-image module.(42:01) Parul unpacked her tutorial "Predicting the Future with Facebook’s Prophet” - a forecasting model to predict the number of views for her Medium articles.(44:58) Parul have been working as a Data Science Evangelist at H2O.AI since July 2019.(47:04) Parul described Flow - H2O's web-based interface (Read her tutorial here).(49:23) Parul described Driverless AI - H2O’s product that automates the challenging and repetitive tasks in applied data science (Read her tutorial here).(52:39) Parul described AutoML - H2O's automation of the end-to-end process of applying ML to real-world problems (Read her tutorial here).(57:07) Parul shared her secret sauce for effective data visualization and storytelling, as illustrated in her analysis of the 2019 Kaggle Survey to figure out women’s representation in machine learning and data science.(01:02:02) Parul described the data science community in Hyderabad, from her lens as an organizer for the Hyderabad Chapter of the Women in Machine Learning and Data Science.(01:05:45) Parul was recognized as a LinkedIn’s Top Voices 2019 in the Software Development category.(01:10:30) Closing segment.Her Contact Info:MediumGitHubTwitterLinkedInWebsiteKaggleHer Recommended Resources:Interpretable Machine Learning post"Interpretable Machine Learning: A Guide for Making Black Box Models Explainable" by Chris Molnar“Towards A Rigorous Science of Interpretable Machine Learning” by Finale Doshi-Velez and Been KimParul’s Compilation of Data Visualization articlesParul’s Programming with Python articlesWomen in Machine Learning and Data ScienceRachel ThomasAndreas MuellerHuggingFace“Factfulness” by Hans Rosling

Show Notes:(2:12) Parul talked about her educational background, studying Electrical Engineering at the National Institute of Technology, Hamirpur.(3:18) Parul worked as a Business Analyst at Tata Power India for 7 years.(4:29) Parul talked about her initial interests in writing about data science and machine learning on Medium.(6:30) Parul discussed her first blog series “A Guide to Machine Learning in R for Beginners” - which covers the Fundamentals of ML, Intro to R, Distributions and EDA in R, Linear Regression, Logistic Regression, and Decision Trees.(8:02) Reference to her articles on data visualization, Parul talked about matplotlib, seaborn, and plotly as the main visualization libraries she practices, in addition to Tableau for building dashboard.(10:11) Parul shared her thoughts on the state of Machine Learning interpretability, in reference to her articles on this topic.(13:54) Parul discussed the advantages of using Jupyter Lab over Jupyter Notebook.(17:30) Parul discussed the common challenges of bringing recommendation systems from prototype into production (Read her two articles about recommendation systems: (1) an overview of different approaches and (2) an overview of the process of designing and building a recommendation system pipeline)(21:00) Parul went in depth into her NLP project called "Building a Simple Chatbot from Scratch in Python (using NLTK).”(23:26) Parul continued this chatbot project with a 2-part series on building a conversational chatbot with Rasa stack and Python and deploying it on Slack.(28:15) Parul went over her Satellite Imagery Analysis with Python piece, which examines the vegetation cover of a region with the help of satellite data.(32:22) Parul talked about the process of Recreating Gapminder in Tableau: A Humble Tribute to Hans Rosling.(35:17) Parul discussed her project Music Genre Classification, which shows how to analyze an audio/music signal in Python.(39:20) Parul went over her tutorials on Computer Vision: (1) Face Detection with Python using OpenCV and (2) Image Segmentation with Python’s scikit-image module.(42:01) Parul unpacked her tutorial "Predicting the Future with Facebook’s Prophet” - a forecasting model to predict the number of views for her Medium articles.(44:58) Parul have been working as a Data Science Evangelist at H2O.AI since July 2019.(47:04) Parul described Flow - H2O's web-based interface (Read her tutorial here).(49:23) Parul described Driverless AI - H2O’s product that automates the challenging and repetitive tasks in applied data science (Read her tutorial here).(52:39) Parul described AutoML - H2O's automation of the end-to-end process of applying ML to real-world problems (Read her tutorial here).(57:07) Parul shared her secret sauce for effective data visualization and storytelling, as illustrated in her analysis of the 2019 Kaggle Survey to figure out women’s representation in machine learning and data science.(01:02:02) Parul described the data science community in Hyderabad, from her lens as an organizer for the Hyderabad Chapter of the Women in Machine Learning and Data Science.(01:05:45) Parul was recognized as a LinkedIn’s Top Voices 2019 in the Software Development category.(01:10:30) Closing segment.Her Contact Info:MediumGitHubTwitterLinkedInWebsiteKaggleHer Recommended Resources:Interpretable Machine Learning post"Interpretable Machine Learning: A Guide for Making Black Box Models Explainable" by Chris Molnar“Towards A Rigorous Science of Interpretable Machine Learning” by Finale Doshi-Velez and Been KimParul’s Compilation of Data Visualization articlesParul’s Programming with Python articlesWomen in Machine Learning and Data ScienceRachel ThomasAndreas MuellerHuggingFace“Factfulness” by Hans Rosling

Show Notes:(2:08) Mael recalled his experience getting a Bachelor of Science Degree in Economics from HEC Lausanne in Switzerland.(4:47) Mael discussed his experience co-founding Wanago, which is the world’s first van acquisition and conversion crowdfunding platform.(9:48) Mael talked about his decision to pursue a Master’s degree in Actuarial Science, also at HEC Lausanne.(11:51) Mael talked about his teaching assistantships experience for courses in Corporate and Public Finance.(13:30) Mael talked about his 6-month internship at Vaudoise Assurances, in which he focused on an individual non-life product pricing.(16:26) Mael gave his insights on the state of adopting new tools in the actuarial science space.(18:12) Mael briefly went over his decision to do a Post Master’s program in Big Data at Telecom Paris, which focuses on statistics, machine learning, deep learning, reinforcement learning, and programming.(20:51) Mael explained the end-to-end process of a deep learning research project for the French employment center on multi-modal emotion recognition, where his team delivered state-of-the-art models in text, sound, and video processing for sentiment analysis (check out the GitHub repo).(26:12) Mael talked about his 6-month part-time internship doing Natural Language Processing for Veamly, a productivity app for engineers.(28:58) Mael talked about his involvement with VIVADATA, a specialized AI programming school in Paris, as a machine learning instructor.(34:18) Mael discussed his current responsibilities at Anasen, a Paris-based startup backed by Y Combinator back in 2017.(38:12) Mael talked about his interest in machine learning for healthcare, and his goal to pursue a Ph.D. degree.(40:00) Mael provided a neat summary on current state of data engineering technologies, referring to his list of in-depth Data Engineering Articles.(42:36) Mael discussed his NoSQL Big Data Project, in which he built a Cassandra architecture for the GDELT database.(47:38) Mael talked about his generic process of writing technical content (check out his Machine Learning Tutorials GitHub Repo).(52:50) Mael discussed 2 machine learning projects that I personally found to be very interesting: (1) a Language Recognition App built using Markov Chains and likelihood decoding algorithms, and (2) the Data Visualization of French traffic accidents database built with D3, Python, Flask, and Altair.(56:13) Mael discussed his resources to learn deep learning (check out his Deep Learning articles on the theory of deep learning, different architectures of deep neural networks, and the applications in Natural Language Processing / Computer Vision).(57:33) Mael mentioned 2 impressive computer vision projects that he did: (1) a series of face classification algorithms using deep learning architectures, and (2) face detection algorithms using OpenCV.(59:47) Mael moved on to talk about his NLP project fsText, a few-shot learning text classification library on GitHub, using pre-trained embeddings and Siamese networks.(01:03:09) Mael went over applications of Reinforcement Learning that he is excited about (check out his recent Reinforcement Learning Articles).(01:05:14) Mael shared his advice for people who want to get into freelance technical writing.(01:06:47) Mael shared his thoughts on the tech and data community in Paris.(01:07:49) Closing segment.His Contact Info:TwitterWebsiteLinkedInGitHubMediumHis Recommended Resources:Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron CourvillePyImageSearch by Adrian RosebrockStation F Incubator in ParisBenevolentAIEconometrics Data Science: A Predictive Modeling Approach by Francis Diebold

Show Notes:(2:08) Mael recalled his experience getting a Bachelor of Science Degree in Economics from HEC Lausanne in Switzerland.(4:47) Mael discussed his experience co-founding Wanago, which is the world’s first van acquisition and conversion crowdfunding platform.(9:48) Mael talked about his decision to pursue a Master’s degree in Actuarial Science, also at HEC Lausanne.(11:51) Mael talked about his teaching assistantships experience for courses in Corporate and Public Finance.(13:30) Mael talked about his 6-month internship at Vaudoise Assurances, in which he focused on an individual non-life product pricing.(16:26) Mael gave his insights on the state of adopting new tools in the actuarial science space.(18:12) Mael briefly went over his decision to do a Post Master’s program in Big Data at Telecom Paris, which focuses on statistics, machine learning, deep learning, reinforcement learning, and programming.(20:51) Mael explained the end-to-end process of a deep learning research project for the French employment center on multi-modal emotion recognition, where his team delivered state-of-the-art models in text, sound, and video processing for sentiment analysis (check out the GitHub repo).(26:12) Mael talked about his 6-month part-time internship doing Natural Language Processing for Veamly, a productivity app for engineers.(28:58) Mael talked about his involvement with VIVADATA, a specialized AI programming school in Paris, as a machine learning instructor.(34:18) Mael discussed his current responsibilities at Anasen, a Paris-based startup backed by Y Combinator back in 2017.(38:12) Mael talked about his interest in machine learning for healthcare, and his goal to pursue a Ph.D. degree.(40:00) Mael provided a neat summary on current state of data engineering technologies, referring to his list of in-depth Data Engineering Articles.(42:36) Mael discussed his NoSQL Big Data Project, in which he built a Cassandra architecture for the GDELT database.(47:38) Mael talked about his generic process of writing technical content (check out his Machine Learning Tutorials GitHub Repo).(52:50) Mael discussed 2 machine learning projects that I personally found to be very interesting: (1) a Language Recognition App built using Markov Chains and likelihood decoding algorithms, and (2) the Data Visualization of French traffic accidents database built with D3, Python, Flask, and Altair.(56:13) Mael discussed his resources to learn deep learning (check out his Deep Learning articles on the theory of deep learning, different architectures of deep neural networks, and the applications in Natural Language Processing / Computer Vision).(57:33) Mael mentioned 2 impressive computer vision projects that he did: (1) a series of face classification algorithms using deep learning architectures, and (2) face detection algorithms using OpenCV.(59:47) Mael moved on to talk about his NLP project fsText, a few-shot learning text classification library on GitHub, using pre-trained embeddings and Siamese networks.(01:03:09) Mael went over applications of Reinforcement Learning that he is excited about (check out his recent Reinforcement Learning Articles).(01:05:14) Mael shared his advice for people who want to get into freelance technical writing.(01:06:47) Mael shared his thoughts on the tech and data community in Paris.(01:07:49) Closing segment.His Contact Info:TwitterWebsiteLinkedInGitHubMediumHis Recommended Resources:Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron CourvillePyImageSearch by Adrian RosebrockStation F Incubator in ParisBenevolentAIEconometrics Data Science: A Predictive Modeling Approach by Francis Diebold

In molti temi della vita e della tecnologia è utile, se non fondamentale, "conoscere il passato per capire il presente e orientare il futuro". Cit. Tucidide 460-404 a.C.Tale citazione può essere applicata anche alla Computer Vision, a maggior ragione perché molte delle tecniche del passato sono tutt'ora applicate per la risoluzione di problemi di visione. Talvolta vengono persino usate congiuntamente tecniche del passato e tecniche moderne al fine di massimizzarne i pregi e limitarne i difetti.Spesso ci viene detto che il Deep Learning è lo state-of-the-art (SOTA) in tema di Visione Artificiale e che, durante il nostro percorso formativo, dovremmo quasi esclusivamente dedicarci a tale tecnologia tralasciando quella che viene definita la Computer Vision "tradizionale".Pur essendo vero che il Deep Learning è a tutti gli effetti lo state-of-the-art, dovremmo prima di tutto studiare in modo approfondito tutto ciò che ha preceduto tale tecnologia, per almeno due semplici motivi: • solo conoscendo la Computer Vision "tradizionale" possiamo comprendere meglio il Deep Learning, i suoi pregi ed i suoi difetti • talvolta non è possibile utilizzare il Deep Learning per motivi di budget o semplicemente perché risulterebbe overkill per la risoluzione di un problema tendenzialmente sempliceCosì come nella programmazione e nel Machine Learning, da un punto di vista progettuale, anche nella Computer Vision non esiste la tecnologia migliore in assoluto, esistono invece vari possibili approcci al problema di cui solo uno è quello ottimale.In questo video verranno illustrate le principali tappe tecnologiche nel campo della Visione Artificiale che hanno scandito le fasi della ricerca scientifica in questo settore, dalle prime intuizioni degli anni '50 alle moderne reti neurali convoluzionali utilizzate nel Deep Learning ed alle più note funzionalità che tutti utilizziamo quotidianamente.

Santiago Molina es ingeniero electrónico de la Universidad Nacional de Colombia, sede Manizales, con maestría en automatización industrial y énfasis en procesamiento de imágenes. Ha trabajado como investigador en la academia, ingeniero de I+D en diversas empresas de tecnología en Manizales. Es experto en machine learning, diseño electrónico, desarrollo de software e integración de sistemas. Habla inglés y alemán. Notas de la conversación: OpenCV https://opencv.org/ BCI (Brain Computer Interface) Age of Empires https://www.ageofempires.com/games/aoeii/ Datos de contacto: Santiago Molina Giraldo. santiago.molina.g@gmail.com Ilustración de portada hecha por Santiago Valencia - IG:Santiago109 --- Send in a voice message: https://anchor.fm/ias-lab-podcast/message

An airhacks.fm conversation with Pavel Pscheidl (@PavelPscheidl) about: Pentium 1 with 12, 75 MHz, first hello world with 17, Quake 3 friend as programming coach, starting with Java 1.6 at at the university of Hradec Kralove, second "hello world" with Operation Flashpoint, the third "hello world" was a Swing Java application as introduction to object oriented programming, introduction to enterprise Java in the 3rd year at the university, first commercial banking Java EE 6 / WebLogic project in Prague with mobile devices, working full time during the study, the first Java EE project was really successful, 2 month development time, one DTO, nor superfluous layers, using enunciate to generate the REST API, CDI and JAX-RS are a strong foundation, the first beep, fast JSF, CDI and JAX-RS deployments, the first beep, the War of Frameworks, pragmatic Java EE, "no frameworks" project at telco, reverse engineering Java EE, getting questions answered at airhacks.tv, working on PhD and statistics, starting at h2o.ai, h2o is a sillicon valley startup, h2o started as a distributed key-value store with involvement of Cliff Click, machine learning algorithms were introduced on top of distributed cache - the advent of h2o, h2o is an opensource company - see github, Driverless AI is the commercial product, Driverless AI automates cumbersome tasks, all AI heavy lifting is written in Java, h2o provides a custom java.util.Map implementation as distributed cache, random forest is great for outlier detection, the computer vision library openCV, Gradient Boosting Machine (GBM), the opensource airlines dataset, monitoring Java EE request processing queues with GBM, Generalized Linear Model (GLM), GBM vs. GLM, GBM is more explained with the decision tree as output, XGBoost, at h2o XGBoost is written in C and comes with JNI Java interface, XGBoost works well on GPUs, XGBoost is like GBM but optimized for GPUs, Word2vec, Deep Learning (Neural Networks), h2o generates a directly usable archive with the trained model -- and is directly usable in Java, K-Means, k-means will try to find the answer without a teacher, AI is just predictive statistics on steroids, Isolation Random Forest, IRF was designed for outlier detection, and K-Means was not, Naïve Bayes Classifier is rarely used in practice - it assumes no relation between the features, Stacking is the combination of algorithms to improve the results, AutoML: Automatic Machine Learning, AutomML will try to find the right combination of algorithms to match the outcome, h2o provides a set of connectors: csv, JDBC, amazon S3, Google Cloud Storage, applying AI to Java EE logs, the amount of training data depends on the amount of features, for each feature you will need approx. 30 observations, h2o world - the conference, cancer prediction with machine learning, preserving wildlife with AI, using AI for spider categorization Pavel Pscheidl on twitter: @PavelPscheidl, Pavel's blog: pavel.cool

丽莎老师讲机器人之可以感知喜怒哀乐的面部跟踪机器人欢迎收听丽莎老师讲机器人，想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导，找丽莎老师！欢迎添加微信号：153 5359 2068，或搜索微信公众号：我最爱机器人。西班牙巴塞罗那自治大学（简称UAB）的研究人员开发了一种具有情绪检测功能的面部跟踪机器人。开发的机器人的创意很大程度上来源于皮克斯公司的顽皮跳跳灯，对创客场景非常感兴趣，并且也有了多年3D打印研究的基础，因此决定着手制作一个‘宠物机器人'来完成一些有趣的人机交互。当然，让机器人像跳跳灯那样跳来跳去是非常困难的，只能尽量保留其‘宠物玩具'的感觉。采用树莓派来开发机器人，机器人主体为Slant Concepts founder开发的机械手LittleArm 2C。研究人员向Slant Concepts founder请求获取修改机器人手臂的权限，以便让其在末端安装摄像头。自行制作了电子外壳和底座。”机器人的相机可以从左至右“扫视”，捕捉照片并使用OpenCV（计算机视觉应用程序常用的编程函数库）检测其中的人脸。“扫视”进行到照片末端后，机器人会自行将摄像头升高或降低，继续其扫视行为。当机器人发现人脸后，会停止扫描，并检查和分析人脸的停留时间。这确保了机器人的扫描结果不会出现假阳性。如果机器人确认它找到了人脸，就会切换至算法中的‘人脸跟踪'部分，进而让人脸图像保持在视野中心。为了达到这个目的，机器人会根据需要进行平移和倾斜。机器人在追踪过程中会拍下人脸，并将其发送到谷歌的云视觉API。谷歌平台会对图像进行分析，检测图像中人物当前的情绪状态，并将其划分为喜悦、愤怒、悲伤、惊讶或者中性等五种情绪状态。机器人收到分析结果后，会模仿用户当时的情绪状态。机器人还配备了全RGB色域LED环，用于动作辅助。根据情绪检测结果，机器人的扫视行为也会发生变化。如果是喜悦，它就会略微加速扫视；如果是愤怒，它就会在不影响面部识别质量的情况下尽可能快地扫视；如果是悲伤，它就会“耷拉着”扫视。每种情绪模式下，机器人的LED环都会闪烁不同的颜色：黄色和暖色代表喜悦，鲜红色代表愤怒，蓝色和冷色代表悲伤，黄色和绿色的混合色代表惊喜。宠物机器人”有广泛的应用前景：让机器人助理更具互动性和自然性，可使残疾人过上更加自给自足的生活；让机器人能识别用户的情绪，可以使老年人及时获得援助；而对小朋友来说，机器人可以及时检测孩子对活动的兴趣度与参与度，并相应地为其调整难度。

In this episode of the SuperDataScience Podcast, I chat with the founder of PyImageSearch.com, Adrian Rosebrock, who gives us a great overview of the space of computer vision. You will learn what computer vision was in the past, what it is now, and most importantly, what it will be in the future and what you need to prepare for if you're interested in computer vision. You will also learn about OpenCV and how to quickly get started with it as it is one of the most popular libraries and tools for computer vision in the world right now. If you enjoyed this episode, check out show notes, resources, and more at www.superdatascience.com/255

The educational technology company Robolink is coming out with a new robotics platform for teaching kids the fundamentals of programming artificial intelligence (AI). Hansol Hong is the CEO and Founder of San Diego-based Robolink. Hansol discusses their latest AI education product Zümi, which won an award at the 2019 International Consumer Electronics Show for innovation. Hansol also talks about their Kickstarter campaigns, the ethics of AI in self-driving cars, drones, and more. Related to this episode: • Robolink: https://www.robolink.com/ • Kickstarter for Robolink's Zümi: https://www.kickstarter.com/projects/robolink/driving-into-the-world-of-ai-zumi • Rokit Smart Robot kit” https://www.robolink.com/rokit-smart/ • Arduino microcontroller: https://www.arduino.cc/ • CoDrone: https://robolink.myshopify.com/products/codrone • Erector Sets: https://en.wikipedia.org/wiki/Erector_Set • UC San Diego: https://ucsd.edu/ • OLED screens: https://www.oled-info.com/oled-introduction • Herbie the Love Bug: https://www.imdb.com/title/tt0083428/ • Raspberry Pi computer: https://www.raspberrypi.org/ • TensorFlow Keras for training learning models: https://www.tensorflow.org/guide/keras • Three Laws of Robotics from Isaac Asimov: https://en.wikipedia.org/wiki/Three_Laws_of_Robotics • “The truth behind Facebook AI inventing a new language,” an article written by AI CTO Roman Kucera: https://towardsdatascience.com/the-truth-behind-facebook-ai-inventing-a-new-language-37c5d680e5a7 • OpenCV for machine vision: https://opencv.org/ • NGSS = Next Generation Science Standards: https://www.nextgenscience.org/ • Common Core Standards: http://www.corestandards.org/ • Udacity courses on AI: https://www.udacity.com/course/ai-artificial-intelligence-nanodegree--nd898 • MOOC = Massive open online course • CES convention: https://www.ces.tech/ • Best of Innovation Award at CES 2019: https://www.ces.tech/News/Press-Releases/CES-Press-Release.aspx?NodeID=df3a115a-ef2a-4aa5-bde1-333397b96973 Subscribe and find more podcast information at: http://www.k12engineering.net. Support Pios Labs with regular donations on Patreon: https://www.patreon.com/pioslabs, or send one-time contributions by buying us coffee: https://ko-fi.com/pioslabs. Thanks to our donors and listeners for making the show possible. The K12 Engineering Education Podcast is a production of Pios Labs: http://www.pioslabs.com.

Quizás este artículo sea uno de los más complicados que he escrito durante la vida de este blog. Pero hablar de Machine Learning, Inteligencia Artificial y Visión Artificial no es sencillo. Sobre todo si quieres que alguien te entienda.Puedes probar a buscar información técnica de Machine Learning. Si es la primera vez, te sonará a chino (como me pasó a mi). También hay información muy general donde se debate más sobre la ética que sobre la técnica.Todos estos campos tienen infinidad de algoritmos, técnicas y métodos que hacen que fácilmente te pierdas.En este artículo planteo una manera de entender qué es y cómo funciona Machine Learning de una forma sencilla y cotidiana. Para conseguir esto tienes que entender varios conceptos que aprenderás en este artículo.Y aunque la teoría es algo muy importante, cuando realmente se aprende es con la práctica. Por eso te planteo un ejemplo práctico con Visión Artificial, para que veas cómo una máquina aprende.Además, todo lo que veas aquí puede ser fácilmente utilizado con tecnologías libres y de código abierto como Arduino, ESP8266 y Raspberry Pi.Espero que disfrutes de ese artículo. Tómalo con calma, prepara una buena taza de café y adelante.Más información en https://programarfacil.com/blog/vision-artificial/que-es-machine-learning/

https://www.humblebundle.com/books/python-packt-2019-books?partner=thetalkinggeek&charity=1827413

https://www.humblebundle.com/books/python-packt-2019-books?partner=thetalkinggeek&charity=1827413

In this episode I talk to Gary Bradski about the creation of OpenCV, Willow Garage, and how to get around institutional roadblocks. Gary is perhaps best known as the creator of OpenCV - an open source tool that has touched almost every application that involves computer vision - from cat-identifying AI, to strawberry-picking robots, to augmented reality. Gary has been part of Intel Research, Stanford (where he worked on Stanley, the self driving car that won the first DARPA grand challenge), Magic Leap, and started his own Startups. On top of that Gary was early at Willow Garage - a private research lab that produced two huge innovations in robotics: The open source robot operating system and the pr2 robot. Gary has a track record of seeing potential in technologies long before they appear on the hype radar - everything from neural networks to computer vision to self-driving cars.  Key Takeaways Aligning incentives inside of organizations is both essential and hard for innovation. Organizations are incentivized to focus on current product lines instead of Schumpeterian long shots. Gary basically had to do incentive gymnastics to get OpenCV to exist. In research organization there's an inherent tension between pressure to produce and exploration. I love Gary's idea of a slowly decreasing salary. Ambitious projects are still totally dependent on a champion. At the end of the day, it means that every ambitious project has a single point of failure. I wonder if there's a way to change that. Notes Gary on Twitter The Embedded Vision Alliance Video of Stanley winning the DARPA Grand Challenge A short history of Willow Garage  

Rob and Jason are joined by Devon Labrie to discuss his experience learning C++ at Augusta Tech and being a first time attendee at CppCon. Adi is an entrepreneur, speaker, consultant, software architect and a computer vision and machine learning expert with an emphasis on real-time applications. He specializes in building cross-platform, high-performance software combined with high production quality and maintainable code-bases. Adi is the founder of the Core C++ users group in Israel. Having worked on proprietary software for most of his career, his most visible contribution to the world of open-source software is, somewhat ironically, the design of the OpenCV logo. News Common Package specification Modules are not a tooling opportunity Herb Pre-trip report Devon Labrie @labrie_devon Links Augusta Technical College SFML C++ Game Programming Udemy Course Sponsors Backtrace Hosts @robwirving @lefticus    

Rob and Jason are joined by Adi Shavit to discuss his spooky C++ Bestiary Blog post, CppCon talks and an announcement from the Core C++ User Group in Israel. Adi is an entrepreneur, speaker, consultant, software architect and a computer vision and machine learning expert with an emphasis on real-time applications. He specializes in building cross-platform, high-performance software combined with high production quality and maintainable code-bases. Adi is the founder of the Core C++ users group in Israel. Having worked on proprietary software for most of his career, his most visible contribution to the world of open-source software is, somewhat ironically, the design of the OpenCV logo. News What Happens in 2098 with C++? JSON For Modern C++ version 3.3.0 released Meeting C++ 2018 Schedule San Diego Pregame - Optional Choices to Make Adi Shavit @AdiShavit Adi Shavit's Blog Links The C++ Bestiary Core C++ Conference C++ Cryptozoology - A Compendium of Cryptic Characters The Salami Method of Cross Platform Development Sponsors Backtrace Hosts @robwirving @lefticus  

Fotos, memes e ilustraciones, los vemos todo el tiempo. Nos ayudan a transmitir ideas de formas más atractivas. En este episodio cubriremos como es posible manipular las imágenes de forma digital. • Notas del episodio: https://marianoog.com/podcast-algortimos/imagenes/

An Australian developer thought his local police force was spending way too much money on their new license plate scanning system. So he decided to build one himself. Here's how he did this, and how he ended up catching a criminal. Written and read by Tait Brown: https://twitter.com/taitems Original article: https://fcc.im/2iJWWuE Learn to code for free at: https://www.freecodecamp.org Intro music by Vangough: https://fcc.im/2APOG02 Transcript: The Victoria Police are the primary law enforcement agency of Victoria, Australia. With over 16,000 vehicles stolen in Victoria this past year — at a cost of about $170 million — the police department is experimenting with a variety of technology-driven solutions to crackdown on car theft. They call this system BlueNet. To help prevent fraudulent sales of stolen vehicles, there is already a VicRoads web-based service for checking the status of vehicle registrations. The department has also invested in a stationary license plate scanner — a fixed tripod camera which scans passing traffic to automatically identify stolen vehicles. Don’t ask me why, but one afternoon I had the desire to prototype a vehicle-mounted license plate scanner that would automatically notify you if a vehicle had been stolen or was unregistered. Understanding that these individual components existed, I wondered how difficult it would be to wire them together. But it was after a bit of googling that I discovered the Victoria Police had recently undergone a trial of a similar device, and the estimated cost of roll out was somewhere in the vicinity of $86,000,000. One astute commenter pointed out that the $86M cost to fit out 220 vehicles comes in at a rather thirsty $390,909 per vehicle. Surely we can do a bit better than that. The Success Criteria Before getting started, I outlined a few key requirements for product design. Requirement #1: The image processing must be performed locally Streaming live video to a central processing warehouse seemed the least efficient approach to solving this problem. Besides the whopping bill for data traffic, you’re also introducing network latency into a process which may already be quite slow. Although a centralized machine learning algorithm is only going to get more accurate over time, I wanted to learn if an local on-device implementation would be “good enough”. Requirement #2: It must work with low quality images Since I don’t have a Raspberry Pi camera or USB webcam, so I’ll be using dashcam footage — it’s readily available and an ideal source of sample data. As an added bonus, dashcam video represents the overall quality of footage you’d expect from vehicle mounted cameras. Requirement #3: It needs to be built using open source technology Relying upon a proprietary software means you’ll get stung every time you request a change or enhancement — and the stinging will continue for every request made thereafter. Using open source technology is a no-brainer. My solution At a high level, my solution takes an image from a dashcam video, pumps it through an open source license plate recognition system installed locally on the device, queries the registration check service, and then returns the results for display. The data returned to the device installed in the law enforcement vehicle includes the vehicle’s make and model (which it only uses to verify whether the plates have been stolen), the registration status, and any notifications of the vehicle being reported stolen. If that sounds rather simple, it’s because it really is. For example, the image processing can all be handled by the openalpr library. This is really all that’s involved to recognize the characters on a license plate: A Minor Caveat Public access to the VicRoads APIs is not available, so license plate checks occur via web scraping for this prototype. While generally frowned upon — this is a proof of concept and I’m not slamming anyone’s servers. Results I must say I was pleasantly surprised. I expected the open source license plate recognition to be pretty rubbish. Additionally, the image recognition algorithms are probably not optimised for Australian license plates. The solution was able to recognise license plates in a wide field of view. Annotations added for effect. Number plate identified despite reflections and lens distortion. Although, the solution would occasionally have issues with particular letters. A few frames later, the M is correctly identified and at a higher confidence rating. As you can see in the above two images, processing the image a couple of frames later jumped from a confidence rating of 87% to a hair over 91%. I’m confident, pardon the pun, that the accuracy could be improved by increasing the sample rate, and then sorting by the highest confidence rating. Alternatively a threshold could be set that only accepts a confidence of greater than 90% before going on to validate the registration number. Those are very straight forward code-first fixes, and don’t preclude the training of the license plate recognition software with a local data set. The $86,000,000 Question To be fair, I have absolutely no clue what the $86M figure includes — nor can I speak to the accuracy of my open source tool with no localized training vs. the pilot BlueNet system. I would expect part of that budget includes the replacement of several legacy databases and software applications to support the high frequency, low latency querying of license plates several times per second, per vehicle. On the other hand, the cost of ~$391k per vehicle seems pretty rich — especially if the BlueNet isn’t particularly accurate and there are no large scale IT projects to decommission or upgrade dependent systems. Future Applications While it’s easy to get caught up in the Orwellian nature of an “always on” network of license plate snitchers, there are many positive applications of this technology. Imagine a passive system scanning fellow motorists for an abductors car that automatically alerts authorities and family members to their current location and direction. Teslas vehicles are already brimming with cameras and sensors with the ability to receive OTA updates — imagine turning these into a fleet of virtual good samaritans. Ubers and Lyft drivers could also be outfitted with these devices to dramatically increase the coverage area. Using open source technology and existing components, it seems possible to offer a solution that provides a much higher rate of return — for an investment much less than $86M. Remember the $86 million license plate scanner I replicated? I caught someone with it. A few weeks ago, I published what I thought at the time was a fairly innocuous article: How I replicated an $86 million project in 57 lines of code. I’ll admit — it was a rather click-bait claim. I was essentially saying that I’d reproduced the same license plate scanning and validating technology that the police in Victoria, Australia had just paid $86 million for. Since then, the reactions have been overwhelming. My article received over 100,000 hits in the first day, and at last glance sits somewhere around 450,000. I’ve been invited to speak on local radio talk shows and at a conference in California. I think someone may have misread Victoria, AU as Victoria, BC. Although I politely declined these offers, I have met for coffee with various local developers and big name firms alike. It’s been incredibly exciting. Most readers saw it for what it was: a proof of concept to spark discussion about the use of open source technology, government spending, and one man’s desire to build cool stuff from his couch. Pedants have pointed out the lack of training, support, and usual enterprise IT cost padders, but it’s not worth anyone’s time exploring these. I’d rather spend this post looking at my results and how others can go about shoring up their own accuracy. Before we get too deep into the results, I’d like to go over one thing that I feel was lost in the original post. The concept for this project started completely separate from the $86 million BlueNet project. It was by no means an attempt to knock it off. It started with the nagging thought that since OpenCV exists and the VicRoads website has license plate checks, there must be a way to combine the two or use something better. It was only when I began my write-up that I stumbled upon BlueNet. While discovering BlueNet and its price tag gave me a great editorial angle, with the code already written. There were bound to be some inconsistencies between the projects. I also believe part of the reason this blew up was the convenient timing of a report on wasteful government IT spending in Australia. The Federal Government’s IT bill has shot up from $5.9 billion to $10 billion, and it delivered dubious value for that blow out. Media researchers who contacted me were quick to link the two, but this is not something I am quick to encourage. A Disclaimer In the spirit of transparency, I must declare something that was also missing from the original post. My previous employer delivered smaller (less than $1 million) IT projects for Victoria Police and other state bodies. As a result, I’ve undergone police checks and completed the forms required to become a VicPol contractor. This may imply I have an axe to grind or have some specific insider knowledge, but instead I am proud of the projects we delivered. They were both on time and on budget. Visualizing the Results The following is a video representation of my results, composited in After Effects for a bit of fun. I recorded various test footage, and this was the most successful clip. I will go into detail about ideal camera setups, detection regions, and more after the video. It will help you better understand what made this iPhone video I took from through the windscreen a better video than a Contour HD angled out the side window. An Ethical Dilemma If you saw the hero graphic of this article or watched the video above, you may have noticed a very interesting development: I caught someone. Specifically, I caught someone driving a vehicle with a canceled registration from 2016. This could have happened for many reasons, the most innocent of which is a dodgy resale practice. Occasionally, when the private sale of a vehicle is not done by the book, the buyer and seller may not complete an official transfer of registration. This saves the buyer hundreds of dollars, but the vehicle is still registered to the seller. It’s not unheard of for a seller to then cancel the registration and receive an ad hoc refund of remaining months, also worth hundreds of dollars. Alternatively, the driver of the vehicle could well be the criminal we suspect that they are. So, although I jokingly named the project plate-snitch when I set it up on my computer, I’m now faced with the conundrum of whether to report what I saw. Ultimately, the driver was detected using a prototype of a police-only device. But driving on a 2016 registration (canceled, not expired) is a very deliberate move. Hmm. Back to the Results Of the many reactions to my article, a significant amount were quite literal and dubious. Since I said I replicated the software, they asserted that I must have a support center, warranties, and training manuals. One even attempted to replicate my results and hit the inevitable roadblocks of image quality and source material. Because of this, some implied that I cherry-picked my source images. To that I can only say, “Well, duh.” When I built my initial proof of concept (again, focusing on validating an idea, not replicating BlueNet), I used a small sample set of less than ten images. Since camera setup is one of, if not the most, important factors in ALPR, I selected them for ideal characteristics that enhance recognition. At the end of the day, it is very simple to take a fragile proof of concept and break it. The true innovation and challenge comes from taking a proof of concept, and making it work. Throughout my professional career, many senior developers have told me that things can’t be done or at least can’t be done in a timely manner. Sometimes they were right. Often, they were just risk averse. “Nothing is impossible until it is proven to be.” Many people bastardize this quote, and you may have seen or heard one of it’s incarnations before. To me, it neatly summarizes a healthy development mindset, in which spiking and validating ideas is almost mandatory to understanding them. Optimal ALPR Camera Setups This project is so exciting and different for me because it has a clear success metric — whether the software recognizes the plate. This can only happen with a combination of hardware, software, and networking solutions. After posting my original article, people who sell ALPR cameras quickly offered advice. Optical Zoom The most obvious solution in hindsight is the use of an optical zoom. Though I explore other important factors below, none lead to such a sheer increase in recognition as this. In general, professional ALPR solutions are offset at an angle, trained on where the license plate will be, and zoomed into the area to maximize clarity. This means the more zoom, more pixels to play with. All the cameras I had at my disposal were of a fixed lens. They included: A Contour HD action camera. These came out in 2009, and I use mine to record my cycling commute and to replay each week’s near death experience. The featured test run was recorded on my phone. My only method of replicating an optical zoom was using an app to record at 3K instead of 1080p, and then digitally zooming and cropping. Again, more pixels to play with. Angle & Positioning The viewing angle of 30° is often referenced as the standard for ideal plate recognition. This is incredibly important when you learn that BlueNet uses an array of cameras. It also makes sense when you consider what a front facing camera would generally see — not very much. What a front facing ALPR camera sees — not much. If I had to guess I’d say a mostly forward-facing array would be the ideal setup. It would consist of a single camera pointed dead center as above, two off-center at 30° each side, and a single rear-facing camera. The value in having most of the cameras pointed forward would come from the increased reaction time if the vehicle is traveling in the opposite direction. This would allow a quicker scan, process, and U-turn than if the rear facing cameras picked up a suspect vehicle already ten meters past the police vehicle. A four camera array would need to be angled similar to this. Icons from Freepik. A Gymbal When compositing the video, I thought about stabilizing the footage. Instead I opted to show the bumpy ride for what it was. What you saw was me holding my phone near the windscreen while my wife drove. Check out that rigorous scientific method. Any production-ready version of a vehicle-mounted ALPR needs some form of stabilisation. Not a hand. Frame Rate Both the attempt to replicate my project and my recordings since then explored the same misconception that ALPR sampling frame rate may be linked to success. In my experience, this did nothing but waste cycles. Instead, what is incredibly important is the shutter speed creating clean, crisp footage that feeds well into the algorithm. But I was also testing fairly low-speed footage. At most, two vehicles passing each other in a 60km/h zone created a 120km/h differential. BlueNet, on the other hand, can work up to an alleged 200km/h. As a way of solving this, a colleague suggested object detection and out-of-band processing. Identify a vehicle and draw a bounding box. Wait for it to come into the ideal recognition angle and zoom. Then shoot a burst of photos for asynchronous processing. I looked into using OpenCV (node-opencv) for object recognition, but I found something simpler like face detection, taking anywhere from 600–800ms. Not only less than ideal for my use, but pretty poor in general. Hype-train TensorFlow comes to the rescue. Able to run on-device, there are examples of projects identifying multiple vehicles per frame at an astounding 27.7fps. This version could even expose speed estimations. Legally worthless, but perhaps useful in every day policing (no fps benchmark in readme). To better explain how high-performance vehicle recognition could couple with slower ALPR techniques, I created another video in After Effects. I imagine that the two working hand-in-hand would look something like this: Idea: how vehicle object detection could remove ALPR frame limits by processing asynchronously. Frame Rate vs Shutter Speed A different manifestation of frame rate is largely influenced upon shutter speed, and more specifically, the rolling shutter issues that plague early or low end digital movie recorders. The following is a snapshot from some Contour HD footage. You can see at only 60km/h the rolling shutter issue makes the footage more or less unusable from an ALPR point of view. Adjusting frame rate on both the Contour HD and my iPhone did not result in noticeably less distortion. In theory, a higher shutter speed should produce clearer and crisper images. They’d become increasingly important if you were to chase the 200km/h BlueNet benchmark. Less blur and less rolling shutter distortion would ideally lead to a better read. Open ALPR Version One of the more interesting discoveries was that the node-openalpr version I was using is both out-of-date and not nearly as powerful as their proprietary solution. While an open source requirement was certainly a factor, it was amazing how accurately the cloud version could successfully read frames that I couldn’t even identify a plate on. ALPR Country Training Data I also found that the main node-openalpr package defaults to US country processing with no way of overriding it. You have to pull down someone else’s fork which allows you to then provide an extra country parameter. Slimline Australian plates need their own separate country detection to regular Australian plates? But this doesn’t always help. Using the default US algorithm I was able to produce the most results. Specifying the Australian data set actually halved the number of successful plate reads, and it only managed to find one or two that the US algorithm couldn’t. Providing the separate “Australian Wide Plate” set again halved the count and introduced a single extra plate. There is clearly a lot to be desired when it comes to Australian-based data sets for ALPR, and I think that the sheer number of plate styles available in Victoria is a contributing factor. Good luck with that. Planar Warps Open ALPR comes with one particular tool to reduce the impact of distortion from both the camera angle and rolling shutter issues. Planar warp refers to a method in which coordinates are passed to the library to skew, translate, and rotate an image until it closely resembles a straight-on plate. In my limited testing experience, I wasn’t able to find a planar warp that worked at all speeds. When you consider rolling shutter, it makes sense that the distortion grows relative to vehicle speed. I would imagine feeding accelerometer or GPS speed data as a coefficient might work. Or, you know, get a camera that isn’t completely rubbish. What others are doing in the industry Numerous readers reached out after the last post to share their own experiences and ideas. Perhaps one of the more interesting solutions shared with me was by Auror in New Zealand. They employ fixed ALPR cameras in petrol stations to report on people stealing petrol. This in itself is not particularly new and revolutionary. But when coupled with their network, they can automatically raise an alert when known offenders have returned, or are targeting petrol stations in the area. Independent developers in Israel, South Africa, and Argentina have shown interest in building their own hacked-together versions of BlueNet. Some will probably fare better than others, as places like Israel use a seven digit license plates with no alphabet characters. Key Takeaways There is simply too much that I’ve learned in the last few weeks of dabbling to fit into one post. While there have been plenty of detractors, I really do appreciate the support and knowledge that has been sent my way. There are a lot of challenges you will face in trying to build your own ALPR solution, but thankfully a lot of them are solved problems. To put things in perspective, I’m a designer and front end developer. I’ve spent about ten hours now on footage and code, another eight on video production, and at least another ten on write-ups alone. I’ve achieved what I have by standing on the shoulders of giants. I’m installing libraries built by intelligent people and have leveraged advice from people who sell these cameras for a living. The $86 million question still remains — if you can build a half-arsed solution that does an okay job by standing on the shoulders of giants, how much more money should you pour in to do a really really good job? My solution is not even in the same solar system as the 99.999% accurate scanner that some internet commenters seem to expect. But then again, BlueNet only has to meet a 95% accuracy target. So if $1 million gets you to 80% accuracy, and maybe $10 million gets you to 90% accuracy — when do you stop spending? Furthermore, considering that the technology has proven commercial applications here in Oceania, how much more taxpayer money should be poured into a proprietary, close-sourced solution when local startups could benefit? Australia is supposed to be an “innovation nation” after all.

Elecia gave a talk about machine learning and robotics at the Hackaday July Meetup at SupplyFrame DesignLab (video!) and LA CrashSpace. She gives it again in the podcast while Chris narrates the demos.  Embedded Patreon Embedded show #187: Self Driving Arm is the interview with Professor Patrick Pilarski about machine learning and robotics applied to prosthetic limbs. I have also written more about my machine learning + robot arm on this blog. My code is in github (TyPEpyt). My machine learning board is Nvidia’s Jetson TX2. The Two Days to a Demo is a good starting point. However, if you are new to machine learning, a better and more thorough introduction is the Andrew Ng’s Machine Learning course on Coursera. To try out machine learning, look at Weka Data Mining Software in Java for getting to know your data and OpenIA Gym for understanding reinforcement learning algorithms I use the MeArm for my robot arm. For July 2017, the MeArm kit is on sale at the Hackaday store with the 30% off coupon given at the meetup (or in Embedded #207). Inverse kinematics is a common robotics problem, it took both Wiki and this blog post to give me some understanding. I wasn't sure about the Law of Cosines before starting to play with this so I made a drawing to imprint it into my brain. Robot Operating System (ROS) is the publisher-subscriber architecture and simulation system. (I wrote about ROS on this blog.) To learn about ROS, I read O’Reilly’s Programming Robots with ROS and spent a fair about of time looking at the robots on the ROS wiki page. I am using OpenCV in Python to track the laser. Their official tutorials are an excellent starting point. I recommend Adafruit’s PCA9685 I2C PWM/Servo controller for interfacing the Jetson (or RPi) to the MeArm. Finally, my talk notes and the Hackaday Poster!  

The Greek Tragedy of ADCs Parker With the success of the Raspberry Pi Compute Module LVDS test board. I started putting together the PinHeck REV8 board More Python and OpenCV work (fun!) Have the webcam taking pictures Auto Crops and records all the images of parts Looking into a higher resolution camera, more info on that later Stephen Parts for the filter finally shipped! Started putting together the boards Like the multi colored jumpers Veroboard - Strip Board Mouser Part Number 854-ST2 80 x 100 mm with mounting holes Use a 7/32" drill bit and a jeweler's drill to cut traces and brillow pad to clean up Going to add a tube preamp to the synth output Already have the preamp built. Just need to connect it. Schematic Tritrix speakers and the nutube amp. Kicking up the nutube amp again The speakers are almost built just need to paint and finish 3d printed a plate for soldering the crossover to Pick Of the Week (POW) Nuvoton NAU7802 - found on the Nice Chips Subreddit Precision low-power 24-bit ADC , with an onboard low-noise PGA, onboard oscillator, and a precision 24-bit sigma-delta ADC . Capable of up to 23-bit ENOB (Effective Number Of Bits) performance. SOP-16 or DIP-16 I2C $2.22 in singles Gotcha @10SPS, PGA=1 Rapid Fire Opinion (RFO) At Last, (Almost) A Cellphone With No Batteries! - HackADay University of Washington The first-ever battery free cell phone, able to make calls by scavenging ambient power. Not really a cellphone. Its really a remote handset for a base station. 3.5 microwatts and transmits 31 ft away. Evaluation boards for USB type-C power delivery - Electronics Weekly Rohm has announced USB Power Delivery (USBPD) transmitter/receiver evaluation boards. 15 to 100WTags: 854-ST2, AND!XOR, BM92A21MWV-EVK-001, Click Bait, MacroFab, macrofab engineering podcast, MEP, NAU7802, Nuvoton, OpenCV, pinheck pinball system, Podcast, Python, RPI3 CM, Tritrix Speakers, Tube Preamp, USB Type-C, Veroboard

Reverse Biasing Opamps Parker Been learning Python and OpenCV OpenCV is an open source computer vision library PyImageSearch Setup PyCharm Using Requests to use and get information from the MacroFab API Built a 3D printed chassis for holding a 5MP USB camera Taking pictures of PCBs and removing lens distortion RPI3 CM PCB done Will be testing over the weekend 6 Pack of beer riding on this board! Board does power up correctly. COMPIOT board Opamp blew up :( Symbol on Schematic unlabeled for the power inputs of the opamp Stephen Still Making the three synth modules (two envelops and one filter) Waiting on the CA3146 transistor IC This is an obsolete part Stephen wants to use it for legacy reasons Making some power supplies for tube mics (the same one that we made for JOSH). Have to figure out how to make a 6.3V or 12.6V high current rail along with a 200V to 300V low current rail. Stephen is going to use transformers that I have at my shop. One drops 120VAC to 10VAC and the other raises the 10VAC up to 230VAC The 10VAC rectified can potentially give 12.6VDC. If not I can drop it to 6.3VDC. The 230VAC rectified can become 325VDC. Pick Of the Week (POW) 130-in-1 Electronic Playground from SparkFun With the closure of Radio Shack where would you buy one of these intro kits? $50 Is cheaper then Radio Shack even! Rapid Fire Opinion (RFO) Fail of the Week: Museum Buttons - Hack A Day Industrial buttons are not really kid proof Wood Lichtenberg - Instructables 12kV 35mA - Neon Sign Transformer PSU Baking soda in water as the conductor for electricity Marinate wood in baking soda water mix 342,857.1428 ohms to pull the entire PSU load Qualcomm fingerprint sensor enables virtual home button - Electronics Weekly Qualcomm has developed an ultrasonic fingerprint sensor which can operate under the glass of a mobile device avoiding the need for a mechanical home button. It also detects heartbeat and blood flow. 800 µm of cover glass and up to 650 µm of aluminum Special thanks to whixr over at Tymkrs for the intro and outro!Tags: CA3146, MacroFab, macrofab engineering podcast, MEP, OpenCV, Podcast, PyCharm, Python, Qualcomm, RPI3 CM, Tube Mics, Wood Lichtenberg

Modal Logic Epistemic Modal Logic Doxastic Logic Backus–Naur form Journal paper based on Iain’s PhD thesis Robot or Not? Netflix Prize The entity benefiting most from AI hype is… The government of Anguilla Iain’s computer vision tracking paper SLAM LSD-SLAM There are four giraffee species Deep Learning results on Google Trends Alexnet Imagenet Challenge Generative Visual Manipulation on the Natural Image Manifold Deep Learning skin cancer identification Argumentation Theory Evolved Antenna HTML5 Genetic Cars Getting Started Practical Dev Tweet PyImageSearch - Seam carving with OpenCV, Python, and scikit-image Nvidia DIGITS ROS + Gazebo scikit-learn Python lib scikit-image Python lib Aftershow VR-Ready power supply

Followup Why DIDN’t Larrabee fail? AVX-512 support in Xeon Phi, Knights Landing CPUs supporting AVX-512 New Nvidia GPUS Laptop GTX 1060, 1070, 1080 Deep Learning aimed P4 and P40 iPhone 7 Plus John Gruber’s iPhones 7 review LITTLE cores not used in Geekbench Benchmarks big.LITTLE bug in Mono Nine-levels of depth detection Remours of Intel modems in iPhone 7 Some Sensor Sums On a 7+, the wide is 3.99mm /_f_1.8 on 1/3”, and the tele a 6.6mm /f2.8 on 1/3.2” Angle of view is α = 2arctan( d / 2f), where d is the sensor width. So decrease f, to get the same α you need to keep (d / 2f) constant, so d decreases. Calculate f number like: N = f/D, where f is focal length and D aperture diameter. So double f and 22 the area of glass needed. But f isn’t doubled on the tele, as the sensor is smaller so doesn’t need to double to half the FOV (2x zoom) - see the above point about α. Speculation and deep dive on future ceramic iPhones Nilay Patel’s iPhone 7 review on The Verge Intel buys Movidius Official announcement from Movidius CEO Open CV Mat Hackett (Not Matt Haughey, sorry) describes the OpenCV install procedure The BBC’s Turing Codec Announcement post and benchmarks Project Homepage GitHub Repo Aftershow Email addresses it would be really annoying to give out over the phone Intertec Superbrain Stackexchange question about Office Space company names HP 9000 Superdome on Ebay HP Integrity Superdome 2 on Ebay Asus E3V5 WS Motherboard Asus Support Website H110M-E/M.2 ASUS ROG Strix GeForce® GTX 1080

Im Gespräch mit Daniel Koester am Computer Vision for Human-Computer Interaction Lab (cv:hci) geht es um die Erkennung des freien Weges vor unseren Füßen. Das cv:hci befasst sich mit der Fragestellung, wie Menschen mit Computer oder Robotern interagieren, und wie gerade die Bildverarbeitung dazu beitragen kann. Das Thema lässt sich auch gut mit dem Begriff der Anthropromatik beschreiben, der von Karlsruher Informatikprofessoren als Wissenschaft der Symbiose von Mensch und Maschine geprägt wurde und im Institut für Anthropromatik und Robotik am Karlsruher Institut für Technologie (KIT) http://www.kit.edu/ erforscht und gelebt wird. So wurde der ARMAR Roboter, der elektronische Küchenjunge (Video), ebenfalls am Institut an der Fakultät für Informatik  entwickelt. Schon früh stellte sich heraus, dass die Steuerung von Programmierung von Computern mit höheren Programmiersprachen wie Fortran, BASIC oder Logo durch Anlehnung an die menschliche Sprache große Vorteile gegenüber der Verwendung der Maschinensprache besitzt. Damit liegt das Thema ganz natürlich im Bereich der Informatik ist aber auch gleichzeitig sehr interdisziplinär aufgestellt: Das Team des KaMaRo (Folge im Modellansatz Podcast zum KaMaRo und Probabilistischer Robotik) entwickelt den Roboter in einem Team aus den Disziplinen Maschinenbau, Elektrotechnik und Informatik. Mit der Freiflächenerkennung befasst sich Daniel Koester seit seiner Diplomarbeit, wo er die Frage anging, wie die Kurzstreckennavigation für blinde Personen erleichtert werden kann. Hier besteht eine Herausforderung darin, dass zwischen einer Fußgängernavigation und der Umgebungserfassung mit dem Blindenlangstock eine große informative Lücke besteht. Nach Abschaltung der Selective Availability des GPS liegt die erreichbare Genauigkeit bei mehreren Metern, aber selbst das ist nicht immer ausreichend. Dazu sind Hindernisse und Gefahren, wie Baustellen oder Personen auf dem Weg, natürlich in keiner Karte verzeichnet. Dabei können Informationen von anderen Verkehrsteilnehmern Navigationslösungen deutlich verbessern, wie das Navigationssystem Waze demonstriert. Die Erkennung von freien Flächen ist außer zur Unterstützung in der Fußgängernavigation auch für einige weitere Anwendungen sehr wichtig- so werden diese Techniken auch für Fahrassistenzsysteme in Autos und für die Bewegungssteuerung von Robotern genutzt. Dabei kommen neben der visuellen Erfassung der Umgebung wie bei Mobileye auch weitere Sensoren hinzu: Mit Lidar werden mit Lasern sehr schnell und genau Abstände vermessen, Beispiele sind hier das Google Driverless Car oder auch der KaMaRo. Mit Schall arbeiten Sonor-Systeme sehr robust und sind im Vergleich zu Lidar relativ preisgünstig und werden oft für Einparkhilfe verwendet. Der UltraCane ist beispielsweise ein Blindenstock mit Ultraschallunterstützung und der GuideCane leitet mit Rädern aktiv um Hindernisse herum. Mit Radar werden im Auto beispielsweise Abstandsregelungen und Notbremsassistenten umgesetzt. Die hier betrachtete Freiflächenerkennung soll aber keinesfalls den Langstock ersetzen, sondern das bewährte System möglichst hilfreich ergänzen. Dabei war es ein besonderer Schritt von der Erkennung bekannter und zu erlernenden Objekte abzusehen, sondern für eine größere Robustheit und Stabilität gerade die Abwesenheit von Objekten zu betrachten. Dazu beschränken sich die Arbeiten zunächst auf optische Sensoren, wobei Daniel Koester sich auf die Erfassung mit Stereo-Kamerasystemen konzentriert. Grundsätzlich ermöglicht die Analyse der Parataxe eine dreidimensionale Erfassung der Umgebung- dies ist zwar in gewissem Maße auch mit nur einer Kamera möglicht, die sich bewegt, jedoch mit zwei Kameras in definiertem Abstand wird dies deutlich leichter und genauer. Dies entspricht dem verbreiteten stereoskopischen Sehen von Menschen mit Augenlicht, doch mitunter kommt es zu Situationen, dass Kinder bei einem schwächeren Auge das stereoskopische Sehen nicht erlernen- hier können temporär Augenpflaster zum Einsatz kommen. Zur Rekonstruktion der Tiefenkarte aus einem Stereobild müssen zunächst korrespondierende Bildelemente gefunden werden, deren Parallaxenverschiebung dann die Bildtiefe ergibt. Ein Verfahren dazu ist das Block-Matching auf Epipolarlinien. Für ein gutes Ergebnis sollten die beiden Sensoren der Stereo-Kamera gut kalibriert und die Aufnahmen vor der Analyse rektifiziert sein. Die Zuordnung gleicher Bildelemente kann auch als lokale Kreuzkorrelation gesehen werden. Diese Tiefenrekonstruktion ist auch den menschlichen Augen nachempfunden, denen durch geeignete Wiederholung zufälliger Punkte in einem Bild eine räumliche Szene vorgespielt werden kann. Dieses Prinzip wird beim Stereogrammen oder Single Image Random Dot Stereogram (SIRDS)  ausgenutzt. Weiterhin muss man die Abbildungseigenschaften der Kameras berücksichtigen, damit die Parallaxverschiebungen auf horizontalen Linien bleiben. Ebenso müssen Vignettierungen ausgeglichen werden. Algorithmen, die nur lokale Informationen zur Identifikation von Korrespondenzen verwenden, lassen sich sehr gut parallelisieren und damit auf geeigneter Software beschleunigen. Für größere gleichmäßige Flächen kommen diese Verfahren aber an die Grenzen und müssen durch globale Verfahren ergänzt oder korrigiert werden. Dabei leiden Computer und Algorithmen in gewisser Weise auch an der Menge der Daten: Der Mensch ist ausgezeichnet darin, die Bildinformationen auf das eigentlich Wichtige zu reduzieren, der Computer hat damit aber große Schwierigkeiten. Für den Flowerbox-Testdatensatz (2GB) wurden Videos mit 1600x1200 Pixeln aufgelöste und synchronisierte Kameras in Stereo aufgezeichnet. Beispiele für synchronisierte Stereokamera-Systeme im Consumer-Bereich sind die Bumblebee oder das GoPro 3D-System. Die Kameras wurden leicht nach unten gerichtet an den Oberkörper gehalten und damit Aufnahmen gemacht, die dann zur Berechnung des Disparitätenbildes bzw. der Tiefenkarte verwendet wurden. Ebenso wurden die Videos manuell zu jedem 5. Bild gelabeled, um die tatsächliche Freifläche zur Evaluation als Referenz zu haben. Der Datensatz zeigt das grundsätzliche Problem bei der Aufnahme mit einer Kamera am Körper: Die Bewegung des Menschen lässt die Ausrichtung der Kamera stark variieren, wodurch herkömmliche Verfahren leicht an ihre Grenzen stoßen. Das entwickelte Verfahren bestimmt nun an Hand der Disparitätenkarte die Normalenvektoren für die Bereiche vor der Person. Hier wird ausgenutzt, dass bei der Betrachtung der Disparitätenkarte von unten nach oben auf freien Flächen die Entfernung kontinuierlich zunimmt. Deshalb kann man aus der Steigung bzw. dem Gradienten das Maß der Entfernungszunahme berechnen und damit die Ausrichtung und den auf der Fläche senkrecht stehenden Normalenvektor bestimmen. Die bestimmte Freifläche ist nun der zusammenhängende Bereich, bei denen der Normalenvektor ebenso aufrecht steht, wie bei dem Bereich vor den Füßen. Die Evaluation des Verfahrens erfolgte nun im Vergleich zu den gelabelten Daten aus dem Flowerbox-Datensatz. Dies führt auf eine Vierfeld-Statistik für das Verfahren. Im Ergebnis ergab sich eine korrekte Klassifikation für über 90% der Pixel auf Basis der realistischen Bilddaten. Die veröffentlichte Software ist im Blind and Vision Support System (BVS) integriert, in der erforderliche Module in der Form eine Graphen mit einander verknüpft werden können- bei Bedarf auch parallel. Eine ähnliche aber gleichzeitig deutlich umfassendere Architektur ist das Robot Operation System (ROS), das noch viele weitere Aspekte der Robotersteuerung abdeckt. Eine wichtige Bibliothek, die auch stark verwendet wurde, ist OpenCV, mit der viele Aspekte der Bildverarbeitung sehr effizient umgesetzt werden kann. Die Entwicklung der Hardware, gerade bei Mobilgeräten, lässt hoffen, dass die entwickelten Verfahren sehr bald in Echtzeit durchgeführt werden können: So können aktuelle Smartphones Spiele Software des Amiga Heimcomputers in einem interpretierten Javascript Emulator auf der Amiga Software Library auf Archive.org nahezu in Orginalgeschwindigkeit darstellen. Für die Umsetzung von Assistenzsystemen für blinde und sehgeschädigte Menschen ist aber auch immer der Austausch mit Nutzern erforderlich: So sind Freiflächen für sich für blinde Personen zunächst Bereiche ohne Orientierbarkeit, da es keinen tastbaren Anknüpfungspunkt gibt. Hier müssen entweder digitale Linien erschaffen werden, oder die Navigation sich weiter nahe an fühlbaren Hindernissen orientieren. Am cv:hci ist der Austausch durch das angeschlossene Studienzentrum für sehgeschädigte Studierende (SZS) unmittelbar gegeben, wo entwickelte Technik sich unmittelbar dem Alltagsnutzen stellen muss. Die entwickelte Freiflächenerkennung war nicht nur wissenschaftlich erfolgreich, sondern gewann auch einen Google Faculty Research Award und die Arbeitsgruppe wurde in der Lehre für ihr Praktikum den Best Praktikum Award 2015 ausgezeichnet.Literatur und weiterführende Informationen D.Koester: A Guidance and Obstacle Evasion Software Framework for Visually Impaired People, Diplomarbeit an der Fakultät für Informatik, Karlsruher Institut für Technologie (KIT), 2013.D. Koester, B. Schauerte, R. Stiefelhagen: Accessible Section Detection for Visual Guidance, IEEE International Conference on Multimedia and Expo Workshops (ICMEW), 2013.

En el capítulo de hoy te voy a hablar de cómo podemos capturar el vídeo con Raspberry Pi, para crear nuestro propio sistema de vigilancia. Pero este no es el único proyecto que podemos crear. En el momento que conectamos una cámara a la Raspberry Pi, se nos abre un mundo de oportunidades. Si lo combinamos junto a la Visión Artificial, sacaremos el máximo provecho y rendimiento.Ya te he hablado en otras ocasiones sobre la Visión Artificial, te dejo a continuación todos los artículos y capítulos del podcast.18. Realidad Aumentada67. Big data y visión artificial81. Visión artificial, OpenCV y Python86. Sensor Kinect, inteligencia artificial al alcance de todos¿Por qué aprender visión artificial?Detección de movimiento con OpenCV y PythonNo voy a hacer más hincapié en las ventajas y beneficios que nos ofrece esta ciencia. Hoy vamos a ver un caso práctico donde analizamos un vídeo con Raspberry Pi.Empezaremos por el principio, veremos que es lo que necesitamos para empezar y el coste que supone. Luego continuaremos con la configuración del hardware, Raspberry Pi y cámara. Por último te daré lo básico para empezar. El primer algoritmo que debemos implementar para detectar movimiento con Raspberry Pi, OpenCV y Python.Vídeo con Raspberry PiRaspberry Pi es un ordenador de propósito general de muy bajo coste. Lo que más me gusta de él es su bajo consumo, podemos tenerlo conectado todo el día. Para que te quede más claro, la Raspberry Pi 2 y 3 consumen unos 4 W/hora. Una torre puede consumir hasta 100 veces más pero pongamos que consume 200W/hora, 50 veces más. Imagínate lo que te puedes ahorrar si lo tienes todo el día encendido. Por este motivo, este dispositivo puede ser un gran aliado en nuestras instalaciones y proyectos.Dentro de los posibles usos dentro del análisis de vídeo con Raspberry Pi encontramos algunos como los siguientes:Espejo inteligente con reconocimiento facial.Detector de presencia con grabación de vídeo.Usos posibles con KinectVisión artificial dentro de la robóticaMaterial necesarioVamos a ver ahora que material vamos a necesitar:Raspberry Pi 2 Model B (41,31 €)Tarjeta Micro SD 16GB (7,49€)Cable alimentación Raspberry Pi 5V 2A (6,99€)Cable HDMI (2€)Ratón y teclado USB (18€) o inalámbrico (18,75€)Cámara de Raspberry Pi 5 MP y 1080p (22,62€)No es necesario comprar todo el material. Seguramente tengas una tarjeta Mirco SD, o un teclado y ratón. También puedes utilizar un cargador de móvil para alimentar la Raspberry Pi pero ojo, este debe suministrar 5V y 2A.Una vez que ya tengas todo el material pasamos a hacer la instalación básica.Instalación básica de Raspberry PiHay multitud de tutoriales en Internet muy buenos. Te recomiendo que veas este tutorial de Gabriel Viso de Pitando. La versión que yo utilizo es Raspbian Jessie y con la que seguiré en este capítulo.Una vez que lo tengamos todo instalado, es importante seguir los siguientes pasos para instalar Python + OpenCV. Empezaremos por Python y todos los paquetes necesarios orientados a la programación científica con este lenguaje de programación.

This week Dave and Gunnar talk about kill switches, killer robots, killer apps, and killing products. David Grohl Alley: Mission Accomplished! Bruce Willis is not moving to Dave’s neighborhood Pycon 2018-2019 in Cleveland? SleepBot Sign up for Gunnar’s Newsletter Remember our discussion about data tampering?  Malware scam appears to use GPS data to catch speeding Pennsylvania drivers Meanwhile: Build your own speed trap camera system with a Raspberry Pi and OpenCV MIT Develops Accurate System for Tracking People, Objects via WiFi Enjoy Architecture of Radio SkullConduct: Biometric User Identification on Eyewear Computers Using Bone Conduction Through the Skull KillerDrone = drone + chainsaw Hacker reveals $40 attack that steals $28k police drones from 2km away Android’s unpatched dead device jungle is good for security D&G This Week in Vendor Abandonment: Google reaches into customers’ homes and bricks their gadgets Don’t forget to set your Google Living Will D&G This Week in Vendor Abandonment (Page 2): TSA spent $47,000 on an app that just randomly picks lanes for passengers Speaking of travel: Poor-taste wi-fi name grounds Perth flight D&G Joke Kit of the Week: TSA Lines Causing Frowns? Send in the Clowns! (and Tiny Horses) Websites take control of USB devices: Googlers propose WebUSB API Meanwhile… Half of people plug in USB drives they find in the parking lot Speaking of Google: Google Spaces and Google Allo and Google Duo Red Hat Product Security Risk Report: 2015 Red Hat Summit Government Luncheon and Red Hat Summit Government Breakout Fighting Unicorns in the latest issue of NASA’s AeroSpace Frontiers (bottom of page 4) D&G This Week in Product Design: Why do BIC pen caps have holes in them? The part of the show where Dave causes Gunnar to have a seizure: The perfect suffix for your “cyber-” buzzword Cutting Room Floor Containerizing Ave Maria Barbershop solo? A Metal Tribute to iPhone Ringtones Arduino-powered LEGO nuclear reactor Bistrobot: Sandwich making robot Milk-based 3D scanner Your Amazon Echo can now teach you all about The Art of War PAC-MAN suit The Swedish Number: Get connected to a random Swede and talk about anything Simulate electronics circuits in your browser w/HTML5 CIA Operative’s 9-Step Hotel Safety Checklist Terrifying Workplace Safety Video Is a 4-Minute Horror Movie Chinese opsec funnies: your foreign boyfriend is a western spy! ShadyURL: Don’t just shorten your URL, make it suspicious and frightening. Super 8 hotel art gallery Hopefully, You’ll Never Need to Use Polycom’s High-Tech Prison Phone Classic Programmer Paintings Creating sparklines in unicode (h/t emorisse) Visual search engine for satellite imagery (h/t emorisse) We Give Thanks The D&G Show Slack Clubhouse for the discussion topics!

Hoy es un capítulo especial donde haré el sorteo del Arduino MKR1000 entre todos los suscriptores del Campus de Programarfacil. Si en el podcast hablo de programación, electrónica, proyectos DIY y todo aquello que seamos capaces de programar y desmontar, en el Campus te doy las pautas paso a paso para que tú puedas hacerlo por ti mismo.Si quieres aprender y profundizar en estas disciplinas, entra en el Campus donde encontrarás cursos de programación, electrónica y proyectos DIY. Además tendrás acceso a un formulario de soporte prioritario que yo mismo contesto. Me puedes enviar cualquier duda o sugerencia. Además, como hoy, haré sorteos del material que vaya comprando gracias a los enlaces de afiliados.¿Por qué hay sorteos?Yo utilizo enlaces de afiliados de Amazon, pero ¿qué son estos enlaces de afiliados? Amazon tiene un servicio por el cual tu puedes poner enlaces a sus productos a cambio de un porcentaje de la venta. Esto no repercute en el precio final. Los productos cuestan lo mismo ya sea por afiliación o por su plataforma.Normalmente este porcentaje es muy bajo, en algunas ventas apenas son unos céntimos por lo tanto, mi decisión es que todo lo recaudado por este medio irá destinado a una bolsa para comprar material que luego se sorteará entre los suscriptores del Campus.Si quieres saber más visita la página del sorteo y mira las estadísticas de compras.SorteoVamos a lo importante, el ganador del sorteo del Arduino MKR1000 es ............. Cristian Gil Rodríguez, enhorabuena y que disfrutes tu nueva placa.Lecciones nuevas publicadas esta semanaComo cada semana voy publicando clases nuevas sobre los cursos activos. Esta semana le ha tocado a:Scratch desde CeroLección 5: estructuras de control en Scratch, condicionales.Introducción a la visión artificial con OpenCVLección 3: fundamentos de la imagen digitalMis proyectosComo ya sabrás, soy uno de los tantos emprendedores que nos estamos buscando la vida. A mi es algo que me encanta, de momento no puedo decir que vaya a vivir de ello pero si que me lo estoy pasando muy bien.Esta semana he estado viendo un tema muy interesante que me comentó José de Logroño, un oyente veterano del podcast y un gran seguidor. Me mandó un vídeo donde podía ver como interactúa Kinect Xbox con Processing. Me quedé impresionado y saqué mi Kinect de la caja y me puse a trastear.Conseguí programar con Processing para que me mostrara la placa, escribiré un artículo las próximas semanas sobre el tema. Pero me quedé atascado con Python, tenía ganas de probarlo con OpenCV. Seguiré intentándolo :).También sigo avanzando en el proyecto del robot Rover. Quiero conseguir hacer un proyecto donde este robot sea totalmente autónomo. Seguiré investigando y cuando tenga material suficiente publicaré un artículo y crearé un curso.Hasta aquí el mini programa de hoy. Muchas gracias a todos por vuestra atención, os deseo que paséis un muy buen día. Nos escuchamos y leemos pronto, un saludo de vuestro amigo Luis del Valle.

Seguramente el tema del que te voy a hablar hoy en el podcast te suene a ciencia ficción y creas que es algo que solo lo podemos ver en las películas del mismo género. Sin duda alguna, no estamos en lo más alto en la gráfica de desarrollo en cuanto a soluciones y aplicaciones en esta materia, pero esto no quiere decir que no podamos investigar y aprender de esta ciencia. Ya te conté ¿por qué debemos aprender visión artificial? y hoy te voy a hablar como podemos introducirnos en la visión artificial, OpenCV y Phyton.Antes de continuar quiero hablarte del Campus de Programarfacil. Si quieres crear tus propios proyectos con Arduino o algún dispositivo Open Hardware, debes dominar dos disciplinas, la programación y la electrónica. En el Campus estoy volcando todo mi conocimiento en estas materias con cursos de diferentes niveles, básico, intermedio y avanzado. Tendrás a tu disposición un formulario de soporte premium y sorteos de material electrónico e informático. Entra y busca tu curso.Este tema no es nuevo en el podcast. Ya he hablado en diferentes capítulos:18. Realidad aumentada44. Tratamiento de imágenes con JavaScript64. Proyectos curiosos con Arduino67. Big Data y visión artificialHoy voy a profundizar en la materia y te voy a dar los pasos necesarios para empezar a programar con la biblioteca más famosa de visión artificial, OpenCV.¿Qué es OpenCV?OpenCV es una biblioteca libre desarrollada originalmente por Intel. Vio la luz en el año 1999. Escrita originalmente en C/C++, su mejor virtud es que es multiplataforma, se puede ejecutar en diferentes sistemas operativos (Linux, Windows, Mac OS X, Android e iOS). También la podemos utilizar en diferentes lenguajes de programación como Java, Objective C, Python y mi favorito C#. Precisamente para este último existe una versión que se llama EmguCV.En junio de 2015 se produjo un hito importante, por fin la versión 3.0 estaba disponible. Si hechas números, en 16 años (de 1999 a 2015) solo ha habido 3 versiones. Esto es debido a que desde un principio esta biblioteca ha sido robusta y muy eficiente.En esta última versión cabe destacar que por fin es compatible con la última versión de Python, la 3.0. Esto permite aprovechar todas las ventajas de la última versión de este lenguaje.Quizás sea la biblioteca de visión artificial más importante y más usada del mundo. Es utilizada por universidades, empresas y gente del movimiento Maker para dar rienda suelta a su imaginación al tratarse de un software libre.Pasos para instalar OpenCV y PythonTe preguntarás ¿por qué Python? Aunque todavía no he tratado este lenguaje de programación ni en el blog, ni en el podcast, si que te puedo contar que Python es muy sencillo de usar, favoreciendo el código legible gracias a su sintaxis sencilla.Debemos ser conscientes que el lenguaje nativo de OpenCV es C/C++, con la complejidad que ello conlleva si queremos utilizar esta biblioteca en nuestros proyectos.Lo que más me gusta de Python es que es un lenguaje fácilmente portable a otras plataformas entre las que se incluye Raspberry Pi. Si además disponemos de una cámara conectada, imagínate lo que podemos llegar a conseguir.Aunque en mi día a día yo utilizo Windows y en el Campus he decidido empezar a con este sistema operativo, se puede hacer de igual manera con Linux y OS X.La decisión de empezar por Windows es muy sencilla. Es el sistema operativo más utilizado del mundo y no porque lo diga yo, solo tienes que ver los datos estadísticos que nos proporciona Net Market Share. Según esta empresa, más del 90% de usuarios utilizan Windows.estadistica-uso-sistema-operativoAún así podemos pensar que es una estrategia de ventas y que esta empresa puede pertenecer al magnate de Redmond. Por eso voy a compartir los datos estadísticos obtenidos de Google Analytics sobre el uso de sistemas operativos en esta web osea, vosotros los usuarios.estadistica-analytics-sistema-operativoComo puedes ver hay una diferencia aplastante con el resto de perseguidores. Por eso he optado empezar por Windows, para poder llegar al mayor número de gente y que nadie se sienta excluido.Lo primero que debemos saber antes de empezar con los pasos a seguir para instalar OpenCV y Python, es que esto ya no es una tecnología plug and play. Estamos acostumbrados a hablar de Processing, Arduino, Scratch y las tecnologías fáciles de usar. Con OpenCV la cosa se complica, sobre todo a la hora de preparar el sistema. Pero yo te voy a dar los pasos necesarios para que empieces de una forma muy sencilla. La instalación consta de 3 pasos.Paso 1: Instalación de Python 3.0 con paquetes adicionalesYa no solo tenemos que instalar el lenguaje de programación, para utilizar OpenCV necesitamos instalar, además, ciertos paquetes de Python que nos hará la vida más fácil cuando desarrollemos aplicaciones en visión artificial.NumPy: es una biblioteca de código abierto que da soporte a vectores y arrays para Python.SciPy: es una biblioteca de código abierto que contiene herramientas y algoritmos matemáticos para Python.Matplotlib: es una biblioteca de código abierto para la generación de gráficos a partir de vectores y arrays.Pip: gestor de paquetes para Python.Se puede instalar cada paquete por separado, pero existen plataformas como Anaconda 3 donde viene todo integrado en un único instalador. Te recomiendo que lo hagas con este tipo de plataformas.Paso 2: Instalar OpenCV para Python 3Quizás este paso pudiera ser el más complicado pero gracias al gestor de paquetes Pip se hace muy sencillo. Solo debemos de descargar la versión para nuestro sistema operativo en formato whl y luego instalarlo. Es muy simple gracias al gestor de paquetes.Paso 3: Instalar el entorno de desarrollo (Opcional)Este paso es opcional, podemos utilizar el bloc de notas de Windows para programar en Python. Mi consejo es utilizar Sublime Text 3 y el plugin Anaconda, que convierte este IDE en un entorno de desarrollo optimizado para Python con todas sus funcionalidades.Y estos serían los 3 pasos recomendados para configurar el sistema. Puedes ir al Campus y ver los como lo hago yo paso a paso con vídeos, imágenes y el código necesario para que todo funcione correctamente.El recurso del oyenteHoy traigo un recurso del oyente especial, el email recibido por Antonio Otero. Ha significado mucho par mi porque el objetivo de este proyecto es precisamente ese, ayudar a la gente y en este caso se ha conseguido.Gracias señores por su buena labor.Siento la necesidad de comentarles una situación. (ya os di las gracias en un comentario, pero quiero extenderme mas)Aparte de mi trabajo como desarrollador web, soy formador de inserción para el empleo. Este año me a tocado dar un curso de microsistemas a un grupo algo especial. (Jóvenes entre 18 y 22 años que digamos andan un poco perdidos por no decir nada mas, unos panoramas....).Acostumbrado a mis clases habituales (para "adultos"), no daba con la manera de interesarles en la materia. El temario es muy variado, SO, hardware, electrónica muy básica, scripts mantenimiento.... todo básico pero muy amplio.No encontraba la manera y estaba sufriendo porque no conseguía enderezarlos, estando al borde de la expulsión de algunos alumnos.El caso es que como todas las mañanas en mi hora de trafico hacia el curso y harto de las noticias de política, se me ocurrió poner vuestros podcasts, y habéis sido una inspiración para mi. Habéis cambiado mi forma de ver algunas cosas, me habéis contagiado vuestra ilusión (quiza ya habia perdido alguna) y como buen virus yo se la he trasmitido a mis alumnos.Con scratch he conseguido que se interesen por la programación, y ahora me hacen script de linux bastante majos. incluso hemos estado con ensamblador (muy básico). Pero espero que me programen arduino con c :-)Con arduino están emocionados (he comprado 4 de mi bolsillo pues el centro no los pone). y eso que aún no los han tocado, pero aprenden la teoría con gran interés deseando ponerla en practica :-)En fin, han cambiado de comportamiento completamente, están involucradisimos y no faltan a una clase, y quiero haceros participes de este éxito.La única mala noticia es que estoy llegando al posdcast de esta semana y no se si aguantare a esperar una semana para escucharos de nuevo :-)Gracias, Antonio OteroYa me despido por esta semana, recuerda que nos puedes encontrar en Twitter y Facebook.Cualquier duda o sugerencia en los comentarios de este artículo o a través del formulario de contacto.

안녕하세요. 오늘은 컴퓨터 비전 오픈소스로 매우 유명한 OpenCV의 개요, 사용법 및 ROS(robot operation system)에서 사용할 수 있는 재미있는 예제에 대한 내용을 간단히 이야기 나눠보도록 하겠습니다.Podcast 방송 링크 - 컴퓨터 비전 오픈소스 OpenCV 개요, 사용법 및 예제에 대한 이야기그림. OpenCV overview

Comenzamos el cápítulo 67, vamos a ver temas muy diversos centrados en Big Data y visión artificial. Veremos qué significa este término, algunos ejemplos, la extracción de datos con data mining y cómo las máquinas son capaces de aprender de la información obtenida.Si quieres aportar o preguntar algo, puedes utilizar los comentarios que están debajo de este artículo, estaremos encantados de comenzar un debate al respecto.Lo primero que tenemos que decir es que no somos expertos en este tema, hay mucha literatura escrita, solo pretendemos dar nuestra visión de todo lo que rodea a Big Data. Ya hemos tratado este tema en alguna ocasión en el podcast (10. Cómo afecta el Big Data a nuestras vidas) y en el blog (El Big Data).No es algo nuevo en nuestra sociedad. La obtención de información y su posterior almacenamiento se lleva haciendo desde hace mucho tiempo. Un ejemplo pueden ser los datos públicos estadísticos del INE, existen datos de 1858 por ejemplo, donde aparece información de Cuba, cuando era una colonia española. Desde entonces han sucedido muchas cosas en nuestro país y en el mundo entero, incluso la revolución tecnológica más importante que ha vivido este planeta.Gracias al avance de la tecnología, hoy podemos disfrutar de almacenamiento de datos cada vez más barato, desarrollo y evolución de la tecnología y la ultraconectividad entre dispositivos. Estos avances han permitido que crezca el volumen de información que se almacena (la media de búsquedas de Google por segundo es 40.000) y que su procesamiento sea cada vez más rápido.Además de toda la información que podemos suministrar los humanos, el IoT (Internet de las cosas) jugará un papel crucial en este sector. Imagínate si obtienes información de alguna instalación agropecuaria, con cientos de sensores enviando información cada segundo, con un tratamiento adecuado de la información podrías acondicionar la meteorología a tu antojo. Esto supondría un ahorro drástico en este sector.Big Data es el empleo más prometedor de 2016 en USA según la revista Forbes, y en España será uno de los más demandados, solo tienes que buscar en cualquier portal especializado en el sector del empleo.Pero el Big Data no es algo extraño para nosotros, sin darnos cuenta estamos contribuyendo diariamente enviando información desde nuestros dispositivos móviles, cuando pagamos con una tarjeta, cuando vamos al médico o simplemente cuando estamos hablando por teléfono.De esto trata el Big Data, de almacenar datos para luego poder extraer patrones a través de la minería de datos, patrones que se repiten y que se pueden detectar. La información obtenida es muy diversa, se pueden obtener patrones de consumo, de gatos o delictivos.Arduino tiene mucho que decir en este aspecto. Es el dispositivo que puede enlazar el mundo real con el mundo estadístico, gracias a la ventana al mundo exterior que ofrece a la tecnología por medio de sus pines digitales y analógicos. Por el contrario Raspberry Pi juega el papel de servidor. Al igual que ha ocurrido con la programación, con la electrónica y con la tecnología en general, gracias a Arduino y Raspberry Pi tenemos al alcance de todos poder utilizar el Big Data en nuestras vidas. Un ejemplo podría ser el Picocluser, un cluster de Raspberry Pi a muy bajo coste. Podemos utilizarlo para aprender a trabajar en paralelo con diferentes máquinas y utilizar en nuestros proyectos del Big Data caseros. Incluye software y material didáctico instalado.Algo muy ligado a los datos es el Machine Learning o el aprendizaje de las máquinas. Muy lejos queda lo que nos imaginamos cuando escuchamos estos términos, robots inteligentes capaces de comportarse como personas. El aprendizaje de las máquinas consiste en detectar patrones de conducta a base de explorar cantidades ingentes de datos, cuantos más, mejores patrones y mejores predicciones. Un ejemplo sería los datos estadísticos de un banco, imagínate que están intentando detectar el uso fraudulento de tarjetas. Si tuviéramos una lista de 100 casos de los cuales 5 son fraudes y de esos 5, 4 siguen un patrón, podremos detectar un posible delito si buscamos ese patrón en toda la información que vamos obteniendo del uso de tarjetas. Esta cantidad de muestras no es significativa y tendríamos un margen de error alto pero, si esto lo extrapolas a millones de operaciones con tarjetas que se practican al año, sería fácil sacar un patrón más eficaz, capaz de detectar fraudes con un margen de error muy pequeño.Esta técnica también se utiliza en la visión artificial, hasta día de hoy existen dos algoritmos utilizados en este marco que destacan por encima de otros. Es importante recalcar que no existe el algoritmo perfecto, en las diferentes aplicaciones que podemos encontrar a día de hoy se mezclan diferentes técnicas según cada caso. La visión artificial dista mucho de ser visión humana, el estado del arte de esta ciencia tiene todavía mucho recorrido para evolucionar, seguramente ni tu ni yo lo veremos. Los dos algoritmos más importantes son SIFT (Scale Invariant Featrue Transformation) y SURF (Speed-Up Robust Features). Estos dos algoritmos analizan las imágenes obteniendo los puntos de interés de un objeto. Para poder crear un sistema basado en la visión artificial, en primera instancia debemos someterlo a un periodo de aprendizaje donde se cataloguen objetos para poder relacionar el conjunto de puntos de interés con dichos objetos. Dependerá de la parametrización de cada algoritmo pero, te aseguro que la cantidad de información que podemos obtener es inmensa, cuanta más información del objeto (diferentes ángulos, perspectivas, iluminación, etc...) más fácil será poder detectarlo, pero también más tiempo de cómputo necesitaremos.Esto también forma parte del mundo Big Data, la información obtenida a través de los algoritmos debe ser almacenada y tratada adecuadamente para poder ser utilizada posteriormente. Si quieres empezar con la visión artificial te aconsejo que explores la librería OpenCV, donde encontrarás no solo los algoritmos ya citados, también podrás binarizar imágenes, detección de movimiento, detección de líneas, etc...Por último te vamos a hablar de las herramientas disponibles para el tratamiento de la información. Estas herramientas hacen que la tarea de explorar y extraer información de los datos almacenados sea más rápida y robusta. Te ofrecen una serie de funcionalidades centradas en la estadística y en la minería de datos, facilitando el tratamiento de volúmenes gigantescos de datos. Entre ellas cabe destacar las siguientes:Lenguaje de programación RIBM SPSSMatlabY esto es todo lo que te queríamos contar del Big Data, Data Mining, Machine Learning y Visión Artificial. Queramos o no estamos pagando el precio por usar servicios "gratuitos". Aún así, dudo que ningún día alguna máquina se pueda adelantar a nuestros pensamientos, el ser humano es impredecible, y menos mal, de lo contrario este mundo sería un cochazo.Recurso del díaHadoopHadoop es un software de Apache que te permite el procesamiento distribuido de grandes conjuntos de datos. La gran ventaja de este framework es que podemos utilizar máquinas más simples, que estén distribuidas en diferentes puntos geográficos para que funcionen como un único servidor. Es un software libre que puede ser utilizado en cualquier proyecto para el Big Data. Está programado en Java y al tratarse de un lenguaje mutliplataforma podremos ejecutarlo en cualquier tipo de máquina, incluso tenemos una versión para Raspberry Pi.Muchas gracias a todos por los comentarios y valoraciones que nos hacéis en iVoox, iTunes y en Spreaker, nos dan mucho ánimo para seguir con este proyecto.

We cross borders into the world of physical computing to chat about hardware hacking at House4Hack. Kenneth, Kevin & Len are joined by Toby Kurien (@TobyKurien), one of the early founders of House4Hack, a maker space in Centurion, and chat about Raspberry PI, Arduino, hacking telescopes, sub-orbital flight (ok, not really), home automation and lots more. House4Hack is an open space dedicated to providing enthusiasts with a community and space where they can build physical computing projects and embedded systems. CHANCE TO WIN! Tweet your favorite episode (and mention us) by 15/2 for a chance to get a R256 discount on your DevConf ZA ticket. Follow Toby & House4Hack on the internet: * https://twitter.com/TobyKurien * http://tobykurien.com * http://www.house4hack.co.za * http://groups.google.com/group/house4hack * http://www.meetup.com/house4hack-centurion/ Here are some resources mentioned during the show: * Arduino - http://www.arduino.cc * House4Hack High Altitude Glider Project - http://www.house4hack.co.za/high-altitude-glider-video * House4Hack PiScope - http://www.house4hack.co.za/piscope * OpenCV - http://opencv.org * Make your own smart watch - http://www.instructables.com/id/Make-your-own-smart-watch/ * Desktop aquaponics - http://www.house4hack.co.za/desktop-aquaponics * Getting started with Arduino Book - https://store.arduino.cc/product/B000001 * Raspberry PI 2 - https://www.raspberrypi.org/products/raspberry-pi-2-model-b/ * Arduino UNO - https://www.arduino.cc/en/Main/ArduinoBoardUno * ESP8266 - http://www.esp8266.com * Intel Edison - http://www.intel.com/content/www/us/en/do-it-yourself/edison.html * Intel Galileo - https://www.arduino.cc/en/ArduinoCertified/IntelGalileo * Arduino IDE Built-in examples - https://www.arduino.cc/en/Tutorial/BuiltInExamples * Activating a camera shutter with an Arduino - http://www.martyncurrey.com/activating-the-shutter-release/ * How to photograph water droplets - http://www.photosbykev.com/wordpress/tips-and-trick/water-droplet-photography/ * X10 & Arduino - https://www.arduino.cc/en/Tutorial/X10 * Other home automation protocols explained - http://www.digitaltrends.com/home/zigbee-vs-zwave-vs-insteon-home-automation-protocols-explained/ * Ubuntu MATE - https://ubuntu-mate.org * OpenELEC - http://openelec.tv * KODI - http://kodi.tv * RetroPie - http://blog.petrockblock.com/retropie/ * 7 fantastic RetroPie game stations - http://www.makeuseof.com/tag/7-fantastic-retropie-game-stations-can-build-weekend/ * Turning the Raspberry PI into an FM transmitter - https://github.com/rm-hull/pifm * Raspbian - https://www.raspbian.org * Magpi magazine - https://www.raspberrypi.org/magpi/ * Raspberry PI Zero - https://www.raspberrypi.org/products/pi-zero/ * Node-RED - http://nodered.org Suppliers mentioned (no affiliation) * http://za.rs-online.com * http://www.communica.co.za * http://robotics.org.za * http://www.mantech.co.za * http://www.netram.co.za * http://www.hobbytronics.co.za * https://www.adafruit.com * https://www.sparkfun.com * http://www.dx.com * http://www.banggood.com Other local maker spaces * Binary Space in the Vaal Triangle - http://www.binaryspace.co.za * Makerlabs in Randburg - http://makerlabs.co.za * The MakerSpace in Durban - http://themakerspace.co.za * Maker Station in Cape Town - http://makerstation.co.za And finally our picks Kevin: * Johnny-Five - http://johnny-five.io * Arduino Starter Kit - https://www.arduino.cc/en/Main/ArduinoStarterKit Kenneth: * Lanseria Airport - http://lanseria.co.za * Adafruit Learn Arduino Series - https://learn.adafruit.com/series/learn-arduino Toby: * Hacker News - https://news.ycombinator.com * uMatrix browser addon - https://github.com/gorhill/uMatrix * Detect and disconnect WiFi cameras in that AirBnB you're staying in - https://julianoliver.com/output/log_2015-12-18_14-39 Len: * Lichess - http://en.lichess.org/

Comenzamos este nuevo capítulo donde vamos a hablarte de proyectos curiosos con Arduino. Te vamos a presentar cuatro proyectos que nos han llamado la atención por diferentes razones. Hemos aplicado nuestros propios criterios, según nuestros gustos y hemos seleccionado cuatro de entre todos los que hemos podido ver en Internet. Si tu conoces algún proyecto interesante o curioso nos lo puedes mandar para que hablemos de él en el programa.Si quieres contactar con nosotros lo puedes hacer de diferentes maneras, a través del formulario de contacto, en el e-mail info@programarfacil.com, en Twitter (@programarfacilc) o en Facebook. También puedes mantenerte al día a través de la lista de distribución.PinokioPinokio es un robot en forma de lámpara. Intenta explorar las expresiones y el comportamiento de la programación o computación en robótica. Gracias a los algoritmos y a la electrónica puede ser consciente de su entorno, enfocándose en las personas. Sus autores son Shanshan Zhou, Adam Ben-Dror y Joss Doggett y está construido con los siguientes materiales:Lámpara flexo, utilizan una de IKEAUna webcam, no debe ser profesional6 servomotores1 Arduino1 pcEl comportamiento y los movimientos de Pinokio tiene muchos rasgos parecidos a los de los animales, esto hace que sea entrañable y genere simpatía entre las personas. Recuerda mucho a la lámpara que sale en las películas de Pixar, esa que aplasta la “i”.Realmente la electrónica y la mecánica son relativamente sencillas, solo debemos ensamblar bien los componentes apoyándonos en estructuras auxiliares como soportes, escuadras y demás. Como corazón de la electrónica utiliza Arduino, imagino que será una placa Mega debido a sus mejores prestaciones de computación y potencia. Pero lo realmente complicado es darle vida a la lámpara, la programación.En la fase de programación han utilizado Processing (lenguaje en el que se basa el IDE de Arduino) y C++ con OpenCV (librería de código abierto de visión artificial). Por supuesto que todo esto debe ir acompañado por su ordenador personal y cableado correspondiente. Aunque en la web del proyecto no dicen que tipo de computadora utilizan, podría utilizarse una Raspberry Pi.Detector de contaminación de aireMuchos de vosotros vivís en grandes ciudades, como Madrid, y sufrìs la contaminación con consecuencias como las restricciones de circulación. Este dispositivo está diseñado para proporcionar al usuario un medio rentable para la determinación de la calidad del aire.Se centra en los cinco componentes del Índice de Calidad del Aire de la Agencia de Protección del Medio Ambiente: el ozono, partículas, monóxido de carbono, dióxido de azufre y óxido nitroso. Este dispositivo detecta todos estos contaminantes excepto dióxido de azufre.El dispositivo también incluye un sensor de gas ciudad para alertar al usuario de fugas de gas o la presencia de gases inflamables. Además, se incluye un sensor de temperatura y la humedad ya que estas condiciones pueden afectar al rendimiento de los sensores de gas. A continuación tenéis los materiales que utiliza.Control y energía:Arduino UnoBateria 5VLCD 16x2 RGBSensores:Shinyei PPD42 Detector de partículas (recipiente)MQ-2 Gas SensorMQ-9 Gas SensorMiCS-2714 Gas Sensor (NO2)MiSC-2614 Gas Sensor (Ozone)Keyes DHT11 (Sensor de temperatura y humedad)Ensamblado:Impresora 3D para la cajaPlaca para soldar componentesPila de litio de 5V10 o 15 cables de calibre 24Los sensores utilizados son relativamente baratos y varían en gran medida de componente a componente por lo que necesitan ser calibrados con concentraciones conocidas de los gases de destino. Esta calibración es la que a nuestro entender supone la mayor dificultad para que funcione correctamente. Aunque siempre estarán los datasheets de los sensores para hacer estimaciones.Si añadiéramos un shield de Ethernet o Wifi y múltiples dispositivos de este tipo podrían tejer una red en las ciudades inteligentes para saber en cada momento la calidad de aire de cada zona de nuestra ciudad y actuar en consecuencia.Fish on WheelsSu propio nombre lo indica, se trata de un pez con ruedas, bueno más bien una pecera con ruedas que se mueve con el pez. Es un artilugio muy divertido y fácil de hacer. El objetivo principal de este proyecto, según sus autores, es mostrar que la visión por computador o visión artificial no está limitada a proyectos de automatización para la recopilación de información. Esta tecnología también puede ser utilizada para que los animales, en este caso un pez, puedan interaccionar con el medio. Fish on Wheels muestra esto al permitir que los peces puedan mover su propia pecera, si son lo suficientemente inteligentes como para entender esto.¿Cómo hacer que se mueva la pecera guiado por el pez? Se trata de un proyecto sencillo, consta de un vehículo robot controlado con Arduino. Puedes encontrar varios en el mercado y de diferentes precios. También necesitarás un ordenador o placa lo suficientemente pequeño para que pueda ir encima del robot y que consuman poco. Todos estamos pensando en la misma, una Raspberry Pi (podemos encontrar una por 40€) pero hay otras alternativas como la placa BeagleBoard de Texas Instruments, es algo más cara pero es la que utilizan para este proyecto. Por último necesitamos una webcam estándar, servirá para obtener la información de donde quiere ir el pez.El funcionamiento es simple, con la webcam captan las imágenes del pez desde la perpendicular es decir, la webcam se situará encima de la pecera. Utilizando el contraste del pez con el fondo de la pecera se determina la posición del pez. Esto lo hacemos a través de la Raspberry Pi o BeagleBoard, donde se harán todos los cálculos. En este proyecto también es factible utilizar la librería OpenCV. Posteriormente se enviará esa información a Arduino para que mueva el robot en esa dirección. Todo esto se hace de manera autónoma, por iniciativa del pez. ¿Te imaginas si en vez de un pez lo hacemos con un delfín? Podríamos ver delfines corriendo por los prados :).Hidden light controlEn el mercado existen proyectos muy complejos y completos de control domotico con Arduino. Este ejemplo es muy simple pero a la vez muy original. Se trata de un simple control de luminosidad de una habitación a través de una pequeña estatua de decoración.Al mover la estatua se enciende la tira de leds, si giras hacia la derecha disminuye la intensidad de luz y por el contrario hacia la izquierda se incrementa. Los materiales que utiliza son los siguientes:Un imán (que ponemos en la figura)un sensor magnético de doble eje3 transistores MOSFETSArduino UNOuna tira de ledsCon el imán y sensor magnético construye una brújula para saber la posición de su mando que es la figura. También a través de este sensor es capaz de saber a qué distancia se encuentra el imán y dependiendo de esta, hace que la luz se encienda o se apague. Por lo que nos cuenta el autor, tiene unos pequeños problemas por fluctuación en los cambios pequeños de ángulo por lo que los discrimina. Evidentemente este sistema no es válido para llevarlo a uso cotidiano. Cualquier otro imán que entrara en el radio de acción haría que el sistema ya no funcionara. Pero si que es muy original y con pocos elementos y una red inalámbrica inusual, el campo magnético del imán, consigue un control remoto de la luz de una habitación.Recurso del diaXively Cloud ServicesSe autodefine como una plataforma para un servicio (PaaS) y está construida expresamente para el internet de las cosas (IoT). La plataforma permite publicar los datos de nuestros sensores que podremos monitorizar de forma online y en tiempo real mediante gráficas y widgets. Cada usuario puede gestionar sus datos, permitiendo que estos sean públicos o no, o solo una parte de ellos. Esta filosofía nos permite utilizar los datos de cualquier objeto del IoT conectado a esta plataforma y utilizarlos en nuestro dispositivo.Muchas gracias a todos por los comentarios y valoraciones que nos hacéis en iVoox, iTunes y en Spreaker, nos dan mucho ánimo para seguir con este proyecto.

02:32 - Julian Cheal Introduction Twitter GitHub Blog 02:49 - Julian’s Background with Robots and Drones Arduino AR.Drone 03:32 - NodeCopter Events 04:31 - Traveling with Robots 05:35 - Julian’s Collection and Projects Julian Cheal: Dancing with Robots Raspberry Pi BeagleBone 07:46 - Giving Demos 09:12 - What Makes Robots? Sinon.JS MQTT Protocol 10:21 - Where is IoT (Internet of Things) Heading? Security 13:11 - Programming Languages NodeBots 14:15 - Tools and Protocols The MIDI Protocol Spark Core voodoospark 17:31 - Programming Challenges Around Hardware Hacking Artoo celluloid 18:49 - Barrier to Entry 20:41 - Getting Kids Started Kids Ruby Arduino Starter Kit 22:09 - Wearables EL Wire (Electroluminescent Wire) 23:18 - LEGO Robotics Mindstorms LabVIEW National Instruments 25:01 - Issues with Hardware Hacking 28:22 - Rubyists and Hardware Julian Cheal: Dancing with Robots JRuby Rubinius 29:45 - Interfacing with Humans iBeacon OpenCV 33:27 - [Kickstarter] CHIP - The World's First Nine Dollar Computer 34:01 - Connectivity  Sphero Carin Meier: The Joy of Flying Robots with Clojure @ OSCON 2013 36:55 - More Interesting Projects Aaron Patterson: Using chicken scheme to read sausagebox values Oscilloscope Picks Jacob Kaplan-Moss Keynote @ Pycon 2015 (Jessica) Kobo Aura H20 (Avdi) Liz Abinante: Unicorns Are People, Too (Re-Thinking Soft and Hard Skills) @ Madison+ Ruby 2014 (Coraline) littleBits (Julian) Jewelbots (Julian) Ruby Rogues Episode #156: Hardware Hacking with Julia Grace (Julian) The End of Mr. Y by Scarlett Thomas (Julian)        

02:32 - Julian Cheal Introduction Twitter GitHub Blog 02:49 - Julian’s Background with Robots and Drones Arduino AR.Drone 03:32 - NodeCopter Events 04:31 - Traveling with Robots 05:35 - Julian’s Collection and Projects Julian Cheal: Dancing with Robots Raspberry Pi BeagleBone 07:46 - Giving Demos 09:12 - What Makes Robots? Sinon.JS MQTT Protocol 10:21 - Where is IoT (Internet of Things) Heading? Security 13:11 - Programming Languages NodeBots 14:15 - Tools and Protocols The MIDI Protocol Spark Core voodoospark 17:31 - Programming Challenges Around Hardware Hacking Artoo celluloid 18:49 - Barrier to Entry 20:41 - Getting Kids Started Kids Ruby Arduino Starter Kit 22:09 - Wearables EL Wire (Electroluminescent Wire) 23:18 - LEGO Robotics Mindstorms LabVIEW National Instruments 25:01 - Issues with Hardware Hacking 28:22 - Rubyists and Hardware Julian Cheal: Dancing with Robots JRuby Rubinius 29:45 - Interfacing with Humans iBeacon OpenCV 33:27 - [Kickstarter] CHIP - The World's First Nine Dollar Computer 34:01 - Connectivity  Sphero Carin Meier: The Joy of Flying Robots with Clojure @ OSCON 2013 36:55 - More Interesting Projects Aaron Patterson: Using chicken scheme to read sausagebox values Oscilloscope Picks Jacob Kaplan-Moss Keynote @ Pycon 2015 (Jessica) Kobo Aura H20 (Avdi) Liz Abinante: Unicorns Are People, Too (Re-Thinking Soft and Hard Skills) @ Madison+ Ruby 2014 (Coraline) littleBits (Julian) Jewelbots (Julian) Ruby Rogues Episode #156: Hardware Hacking with Julia Grace (Julian) The End of Mr. Y by Scarlett Thomas (Julian)        

02:32 - Julian Cheal Introduction Twitter GitHub Blog 02:49 - Julian’s Background with Robots and Drones Arduino AR.Drone 03:32 - NodeCopter Events 04:31 - Traveling with Robots 05:35 - Julian’s Collection and Projects Julian Cheal: Dancing with Robots Raspberry Pi BeagleBone 07:46 - Giving Demos 09:12 - What Makes Robots? Sinon.JS MQTT Protocol 10:21 - Where is IoT (Internet of Things) Heading? Security 13:11 - Programming Languages NodeBots 14:15 - Tools and Protocols The MIDI Protocol Spark Core voodoospark 17:31 - Programming Challenges Around Hardware Hacking Artoo celluloid 18:49 - Barrier to Entry 20:41 - Getting Kids Started Kids Ruby Arduino Starter Kit 22:09 - Wearables EL Wire (Electroluminescent Wire) 23:18 - LEGO Robotics Mindstorms LabVIEW National Instruments 25:01 - Issues with Hardware Hacking 28:22 - Rubyists and Hardware Julian Cheal: Dancing with Robots JRuby Rubinius 29:45 - Interfacing with Humans iBeacon OpenCV 33:27 - [Kickstarter] CHIP - The World's First Nine Dollar Computer 34:01 - Connectivity  Sphero Carin Meier: The Joy of Flying Robots with Clojure @ OSCON 2013 36:55 - More Interesting Projects Aaron Patterson: Using chicken scheme to read sausagebox values Oscilloscope Picks Jacob Kaplan-Moss Keynote @ Pycon 2015 (Jessica) Kobo Aura H20 (Avdi) Liz Abinante: Unicorns Are People, Too (Re-Thinking Soft and Hard Skills) @ Madison+ Ruby 2014 (Coraline) littleBits (Julian) Jewelbots (Julian) Ruby Rogues Episode #156: Hardware Hacking with Julia Grace (Julian) The End of Mr. Y by Scarlett Thomas (Julian)        

Noticias y revisión de unos cuantos podcasts sobre Python http://podcast.jcea.es/python/4 Notas: 00:00: Presentación Python en Españo, capítulo 4. 00:47: Hablamos sobre los últimos eventos Python en España del calendario. Python Vigo. 04:40: Samuel y Jesús comentan sobre Python 3.5 Alpha 4 (última alpha). 06:00: Hablamos sobre el anuncio oficial de PyConES 2015 en Valencia. 08:33: Hablamos sobre los vídeos y fotos de la anterior PyConES (2014). 09:40: Las actas. 11:00: Hablamos sobre la ampliación del Call For Proposals de EuroPython 2015. 13:27: Comentamos sobre la reciente publicación de los vídeos de PyCon Americana. 14:00: Jesús Cea nos comenta cosas sobre la última asamblea de la Asociación Python España. 16:45: Las actas. 18:15: Comentamos los últimos podcast sobre Python a nivel mundial que han surgido. 18:55: Jesús nos presenta el podcast "Talk Python To Me". 19:37: Jesús nos presenta el podcast "Podcast.__init__". 20:00: Jesús habla sobre el último podcast de CAChemE sobre Python. 21:20: Jesús habla sobre el último número del podcast de Entre Dev y Ops a raíz de la entrevista de dos organizadores de EuroPython. 22:48: Jesús presenta la última feature de este podcast, shownotes en tu reproductor preferido! 23:45: Juan Ignacio habla sobre el podcast "from Python import Podcast". 24:45: Jesús nos presenta uno de los primero podcast historicos sobre python: "Python 411". 26:51: Juan Ignacio nos habla sobre los tutoriales y post de Adrian Rosebrock sobre OpenCV en su página http://www.pyimagesearch.com/. 31:53: Pedimos feedback, sugerencias, críticas! y hablamos sobre el twitter del podcast. 33:26: Nos despedimos como siempre informado sobre nuestro email y twitter para que contactéis con nosotros.

This week, Dave and Gunnar talk about: robotics, public and not-so-public goods, and redesigning airlines. Subscribe via RSS or iTunes. Adam Clater: please explain. Dave is glad he’s moving to US Airways: United Airlines drops Cleveland as hub airport HT Cleveland expat James Labocki: Did Burke Lakefront Airport Miscalculations Add to Hopkins Hub Troubles in Cleveland? Meanwhile: Delta’s 80’s In-Flight Safety Video Almost related: Dave makes contingency plans for in-flight calling with DrumPants Risacher cited Hellekson’s Law twice at the Adobe Government event last week. Woo! Robotics fun: Team 2399‘s Epic Weekend! This year’s challenge Team 2399’s robot vision code on GitHub with OpenCV Ars rolls in with a telepresence robot Goldfish driving a robot car CFPB pull request of the week: I want RSS, you don’t have RSS, so here’s a patch enabling RSS on your website Comcast customer surprised to learn new router is also public hotspot Graph only Verizon FIOS here: Netflix ISP Speed Index Netflix speeds are down. But don’t blame Verizon. How Netflix and Google Could Lead the Fight For Net Neutrality NBC Olympic Coverage Is Killing Cord Cutters like Gunnar Dave is moderating a Red Hat Summit panel of Innovation Award finalists! HT Langdon White: Check out DevNation April 13-16 while at the Red Hat Summit! Red Hat and Hortonworks Deepen Strategic Alliance on Hadoop New by Dave: 2 questions to ask before diving into infrastructure-as-a-service RedHatGov on GitHub! British Airways RHEVs up private cloud Try it yourself! Boarding pass redesign More boarding pass redesign Ticketmaster redesign Gunnar’s fitbit and stated vs. revealed preferences The Perfect Way to Hold a Hamburger, Proven by Science Cutting Room Floor NSA-o-matic When not denouncing people in North Korea, try the State of the Union generator Postmodernism Generator Scott Pakin’s automatic complaint-letter generator Klingon valentines Evil Mad Scientist Valentines We Give Thanks Dan Risacher for mentioning Hellekson’s Law! Langdon White for reminding us to talk about DevNation!

The theme of today's show is "Super Jumbo" and your hosts are Jason Kridner, Gerald Coley and Jeffery Osier-Mixon.To provide questions or suggestions:Call +1-713-234-0535 orvisit the BeagleCast suggestions formLinks to the recordingsBeagleCast-20110325.mp3BeagleCast-20110325.oggHeadline newsWindows Compact 7 Two new Distributors in China -- ChipSee -- CATCAN From the RSS feed FFmpeg fork becomes libav Clojure on The Beagleboard -- What is a closure vs. what is Clojure?The 2.6.38 kernel is out What Is CLFS? - File System - Check! OpenEmbedded at CeBIT 2011 -- Should we still be excited about CES and CeBIT? Face chasing BeagleBoard-based robot using a Kinect How to build QT Framework 4.7.2 and OpenCV 2.2 for Beagleboard-xM -- How to build sample program for capturing image from camera (OpenCV and Qt) SPI with Trainer-xM Running CyanogenMod on BeagleBoard -- What is CyanogenMod? -- What is Rowboat?NEWS IGEPv2 goes Open Hardware Open Hardware Summit date announced for 2011? -- Looking for votes on a logoBeagleboard: Power usage (current draw) for certain scenarios Upcoming events Indiana Linuxfest OpenSource COM BOF/LUG Mumbai Meeting, 26th March 2011 Community activityGSoC Update -- BeagleBoard.org not a mentoring organization this year -- Still looking for mentors to volunteer to mentor in other projects -- Considering a smaller scale BeagleBoard Summer of Code Always Innovating Announcement...Super Jumbo Beagle Buffet! Upcoming Khasim Syed Mohammed will be on next week to discuss the Android Rowboat project

In this episode we introduce opencv, a popular open source computer vision library created by Dr. Gary Bradski and developed in large part by a team of Russian computer vision and optimization experts. We interviewed Gary at his office at Willow Garage, where they are building an open research platform for personal robotics, including the [...]