3D space motion dense based team tactical status detection in volleyball game analysis

In volleyball game analysis, the team tactical status plays an important role in analyzing game tactics, evaluation of team performance and developing team works for coach. In this paper, the team tactical status is classified into four categories: the defensive ready, the defensive, the offensive ready and the attack. The difficulties to detect one team tactical status from other types including: 1) team rotations and player exchange, 2) different team formations, which make the same team tactical status have various features such as different player position and motion. This paper proposes a 3D space motion dense based team tactical status detection method to solve the complex features of team status. Instead using the local feature of each player, the 3D space motion dense feature describes the team status from two main aspects, the entire team motions relative to the court area and the relative motion of all the players to the ball. With the 3D ball trajectories and multiple players' positions tracked from multi-view volleyball game videos, the experimental result shows the detection accuracy reaches more than 80%.