We want algorithms that can tell us what went where in a video. Tracking is hard because each situation is different, featuring a different camera operator, and subject(s) whose appearance, motion, an
