Content with tag:
Robot Plate Detection Model
As AI team lead of Queens University's Robomaster Design Team, I led twenty students toward learning, designing and creating auto-targeting robot artificial intelligence. Using TensorFlow's Object Detection API and transfer learning on YOLOv3, we successfully deployed a detection algorithm that creates bounding boxes around enemy robot hit plates. Automating this process has given us a considerable advantage against the competition, and we learned a ton in the process! We found that the way we labelled data, and differentiating between red & blue coloured plates had a considerable effect on how accurate our model was.