Object detection with multiple cameras

The paper presents an algorithm for detecting objects from multiple cameras. The algorithm (1) minimizes the effect of shadows, (2) detects and solves occlusion and (3) fuses objects that are detected in more than one camera. The algorithm was developed as part of SORTTS (soccer real-time tracking system), a tracking system based on a set of video cameras that covers a large area. The paper also reports the results of this algorithm in a challenging application area for tracking: professional soccer games.