Hi, I am creating a color tracking interactive installation in which participants activate sound by picking up large primary colored spheres. As I don't want items of clothing, say a red cardie, also triggering the sound I want the shape of the sphere to be giving information as well.
Once the color analysis is done I have looked to use CV.undergrad for shape recognition but I can't work out how to teach it what shapes to look out for. Every time I experiment with the undergrad help patch it makes max crash.
My initial question is...Without being able to experiment I wonder is undergrad the right object even? I have noticed while using other CV.jit objects that as soon as the sphere is picked up the cv objects register person and sphere as one blob, how do I get around this problem?
My second question is ...If it is the right object how do you teach undergrad to compare a still image to the live stream?
I thank you in advance for any help, hints, pointers in the right direction.