Edge computing and machine vision have massive potential to solve real-world problems in manufacturing, logistics, and a wide range of other industries. Some of the best ideas come from innovators in the small and medium-sized business (SMB) space, but developing prototypes and taking solutions to market is challenging without backing and support. 20/20 Vision Hack organizers wanted to inspire solution builders to develop edge-based machine vision systems and provide them with the tools to advance their work – and for one outstanding innovator, the resources to fully develop and market a promising idea.
The competition was hosted by HackerEarth and held from December 15, 2020, through March 31, 2021. ADLINK, in collaboration with Intel, sponsored the event and awarded finalists the Vizi-AI™ Machine Vision Development Kit to provide the solution builders in the competition with a scalable starting point for artificial intelligence (AI) deployments at the edge.
Meet the 20/20 Vision Hack Winner
20/20 Vision Hack finalists submitted a field of excellent submissions – from a gaming engine to worker safety systems. Choosing a winner was a challenge for the event’s judges, but in the final tally, Sergio Velmay emerged as the winner.
Velmay drew upon his experience as a software developer focusing on AI, machine vision, machine learning and chatbots, as well as his background in architecture, 3D design, virtual reality (VR) and drones to take the top prize. He designed his “Machine Vision Poka-Yoke” prototype to give manufacturers a greater ability to monitor and control assembly. The system guides workers through assembly processes, confirming they are using the right parts in the correct number , and it can even detect whether parts are oriented correctly.
In an interview, Velmay told Smart Industry, “I think my proposal stood out in the final process by taking advantage of ADLINK’s Vizi-AI kit device and Intel’s OpenVINO toolkit with a large number of customized deep-learning models.”
The Vizi-AI Machine Vision Development Kit is powered by the Intel Atom x5-E3940 system on chip (SOC) and includes the Intel Movidius Myriad X vision processing unit (VPU) for AI acceleration. The kit also includes the Intel distribution of the OpenVINO™ toolkit, which supports deep learning inference at the Edge, heterogeneous execution across Intel CPUs, Integrated Graphics, Neural Compute Stick 2, and Vision Accelerator Design with Movidius VPUs. The Vizi-AI development kit enables developers to connect to image capture devices and use vision data to deploy and improve machine learning models to optimize operational decision making.
Velmay’s project also stood out to judges because he 3D-printed parts to create a transportable prototype to use during the hackathon to develop datasets and validate his models – and he planned for live, onsite demos.
The Benefits of Competing in a Hackathon
Looking back on the experience of competing in 20/20 Vision Hack, Velmay says, of course, receiving the OpenVINO toolkit, the ADILNK Vizi-AI Machine Vision Development Kit, hardware and the $10,000 first prize are a benefit. In the Smart Industry, however, he commented that there are other reasons to participate, such as “the ability to learn directly from mentors and specialists and the opportunity to be part of a community.”
Interestingly, because 20/20 Vision Hack took place during the COVID-19 pandemic, it followed a virtual format, and instead of detracting from the experience, it enhanced it. The virtual workspace allowed innovators from around the world to participate – no travel required. It also enabled event organizers to provide content and education in real time that related to new technology and tactics. The virtual format also gave participants a chance to meet online and collaborate to build prototypes that now have the potential to result in real solutions in a variety of markets.
Next Steps in Machine Vision Poka-Yoke Development
Although this hackathon has ended, Velmay and the 20/20 Vision Hack are still working to achieve the ultimate goal of advancing the adoption of AI at the edge.
With go-to-market support from ADLINK and Intel, Velmay is preparing to take his system to market. Velmay is working with these 20/20 Vision Hack sponsors on market validation and technical product readiness. Additionally, the ADLINK Edge Solutions team is working with Velmay to create marketing collateral such as a use case, solutions brief, demo video, press release, web page content and a social media campaign. Velmay will also receive support for market validation, campaign execution, and advice on technical product readiness. Exciting things are coming for the Machine Vision Poka-Yoke system and the manufacturers who will benefit from it. Follow ADLINK on social media- Twitter, LinkedIn and Facebook – to watch the story unfold as Velmay takes his product to market!
