has anyone successfully achieved hand-only recognition using cv.jit? The easiest method seems to be to change the model used in cv.jit.faces. Just wondering if anyone has done this, and any useful tips
Changing the cv.jit.faces models should work, but only for one particular view of a particular hand pose. If you just want to recognize any hand, well, you're staring at a very big mountain. If you want to recognize particular hand postures, and you can get a good segmentation of the hand, I managed to very, very accurately recognize up to seven poses using cv.jit.learn.