Closed bubner closed 7 months ago
Discovered a pipeline that can detect Custom Team Prop objects based on averages of tri-sections. TFOD seems to be out of the question at the moment as we don't have a field set up to test.
OpenCV pipeline works well, 3d printing is all done. Ready to go, just need tasks to be done.
Option A) TFOD Use TFOD to train a model for a custom element (red/blue) to pick up on the Spike Marks. This may need to be done after the robot starts and cannot be done during the init-phase (due to potentially having to drive up to the marks)
Option B) OpenCV OpenCV will be faster than TFOD, but rules regulate the colour of objects as team props. This is the preferred option, as we might be able to squeeze this into pre-init, but it's currently unknown.