The Coordinated Activities of Players in Sports (CAPS) project.

We are based at Kingston University, London, England; our research is funded by the Engineering and Physical Sciences Research Council (EPSRC).


Sports recordings provide exciting opportunities for researchers in the fields of Computer Vision and Pattern Recognition. Sports games typically involve player activities that are structured around patterns of play known as tactics and strategies. These are difficult to model because they involve more than one player and typically involve coordinated activities.

The CAPS project aims to produce new techniques to represent and model such tactics and strategies. Our research focuses on the tennis game. The rules of the game help restrict and define the problem: there are only two players; each is restricted to his or her side of the court; and the videos look very similar no matter where the game is played.

insert text
© CAPS Project.

However, the footage provides a gamut of research issues: shadows, occlusions, very fast motion, noise, accurate tracking, and so on. Once we move past these problems, we start a new journey of Pattern Recognition. In a game like tennis, the players engage in known, well-documented patterns of court-moves, body-postures, ball-shots, and so on. How easy is it to model these patterns?