A common problem in vision based HCI
racking the hands, but how to drop
How to get to Buxton s 3 state model
selection
tracking