STOP: Space-Time Occupancy Patterns for 3D Action Recognition from Depth Map Sequences