Some Tips from a Grand Prize Winner of the 2021 OpenCV A.I. Competition
Wow, we just won the 1st place grand prize globally! I can't even begin to describe my excitement when I heard the news. I spent the last few days reflecting on our journey. I want to offer some tips on how to successfully complete your project for future participants and AI enthusiasts.
The most important thing to start is knowing that there is a massive community behind OAK-D. Whenever you have any technical problem, you can always find help in Luxonis' Discord channel. Everyone there is so helpful and friendly. It is one of the best technical communities I have ever joined. I am really impressed by how fast people from Luxonis respond to my questions. When I was exhausted and tired from working on our project, the enthusiasm and vibe from the community consistently managed to uplift my spirit and enabled me to keep going.
Another thing that I found helpful was to look around and learn from the other participants. You can find a lot of information about different projects on Twitter and LinkedIn. OpenCV also sent out team highlights from the competition. This is an excellent way to be inspired by their work. I am seriously amazed by the creativity of the competing teams. There were so many surprising uses of spatial A.I., and I am constantly motivated to re-examine my problem-solving approach. This process often led to epiphany moments, and I ended up significantly simplifying my initial solutions. For example, the spatial Blazepose project by geaxgx led me to think about adding depth to the fall detection algorithm that I was coding. Fortifying body landmarks with depth values from the stereo camera made our fall detection system almost trivial to implement and ten times more robust than before.
Oh, and posting weekly updates on social media was an excellent way to plan our immediate milestones. Meeting these small milestones allowed us to evaluate precisely where we were in our big-picture plan. Our social media posts also engaged the community around us. Their feedbacks and encouragements continually pushed us forward.
To summarize, here are some takeaways for future OpenCV A.I. competition participants.
Join the OAK-D community, participate in discussions, and don't be shy to ask for help.
Join social media platforms, subscribe to the OpenCV newsletter and keep getting inspired by others' work.
Find a way to help yourself to plan ahead and continually evaluate your progress towards finishing your project. Providing weekly social media updates was an excellent way for us to achieve that.
Most importantly, make sure to have a lot of fun and enjoy the moment. Finally, I leave you with an image of the new and improved Charlie, which is now powered by the OAK-D-Lite. Please go and support the OAK-D-Lite's Kickstarter campaign if you haven't already.