Tactic detection in soccer video using ball deviation ratio

Recently, tactic analysis has become an emerging research topic in sports video analysis. Several methods have been proposed to detect the tactics in broadcast soccer videos. In these methods, the tactic is modeled by position labels of the ball or players. As we know, the description ability of the label based feature is lower than the real-number based feature. To address this problem, a novel tactic detection method using the real-number ball deviation ratio is proposed in this paper. In our method, each clip is represented by a sequence of the deviation ratio value, which is the degree of the ball deviates from the vertical middle line of the playfield. Then, the tactic is detected using this real number sequence. Experiment results show the effectiveness of the proposed method.