An adaptive focus-of-attention model for video surveillance and monitoring