This guy is writing:
"I’m sure mine doesn’t work as well as Sony’s, it uses simple image difference to influence the cursor. "
It works with a sort of collision of your hand. It doesnt recognize your hand.
But what are the circumstances and surroundings of your project? Is only the user’s hand visible to camera or the whole body? Are there many people or only one? Is the light sitation changing or is it fixed? Is it important to see every finger or just a hand? Do you wanna move by gestures or by a movements direction?