My team and I are working on a project with the Persee that entails a user pointing at a point in space to then light up an array of LED's. At the current moment, we are having a hard time getting the Persee to recognize the hand location quickly. It seems to be a 4 (or 2, depending on how you look at it) step process: expose the camera to a face, wait for recognition of a face (not a specific face, but that a face is there), expose the camera to a hand, and finally wait for recognition of a hand.
I was wondering if anyone has come up with a way to quickly locate the hand without having to go through these steps? Although these steps work in an entertainment setting, such as using the hand as a remote, it is not especially welcoming to users outside of this environment.