Analyzing Video Sequences of Multiple Humans by