Hackathon
Honors & Awards
honor title München Oerlikon Hacakthon 1st Place Computer Vision honor dateNov 2019 honor issuerOerlikon, München, Deutschland honor description1st place Computer Vision challenge winner and 3rd place overall
Our team of 3 competed against teams of 3-6 over a 24 hour hackathon to create a solution for identifying manufacturing parts in real time using computer vision. We constructed a full physical unit in the makerspace that would allow automation with motors (rather than human assisted as the challenge had originally only required) and connected our computer to the camera attached to the top of the unit which fed us a live feed of items placed in the space. These items were then measured and classified, the quantity and order of the items was random but within a pre-defined scope of parts. We were able to do this with classical computer vision techniques involving OpenCV using edge detection and predetermined scaling due to the unit allow for fixed distances between the camera and the play space. We were given no data to work with and so we chose a solution that didn't require it given the time constraints. We also constructed a user friendly GUI that simply requests the human user (if there was one) to confirm the items and then pull a tray to clear the contents into a bin which could be then taken to storage with a bin we placed under.