This is just a screenshot of my camera view
Hi, the easiest thing to do here is to train a CNN using edge impulse based on the FOMO architecture. It will run fast on your camera and should be able to deal with the lack of contrast.
Is there any guide that I can look at? For my level that sounds pretty complex.
Color Tracking works best where there’s a lot of contrast. Since the ball is solver and the background white… It’s going to be hard with that.
Find circles will struggle since the object changes in size.
