Team formation mapping and sequential ball motion state based event recognition for automatic data volley

Event recognition is an important topic in the volleyball analysis system Data Volley, in which events are classified by their influence to the progress of the game. Normally analysis on Data Volley system relies on entering event data manually but now methods for automatic data acquisition are in demand. This paper proposes a formation mapping and sequential ball motion state based event recognition method for automatic Data Volley system. The team formation mapping method distinguishes those events with similar ball motion by representing the distribution of players when the event happens. Sequential ball motion state feature improves the recognition result by indicating the status of game progress. Experiments are conducted on game videos from the Semifinal and Final Game of 2014 Japan Inter High School Games of Mens Volleyball in Tokyo Metropolitan Gymnasium. Experiments of the proposed method achieve the average accuracy of 98.51% with an improvement of 10.34%, the average recall of 98.94% with an improvement of 18.5% and precision 97.85% with an improvement of 13.12% comparing to the conventional method.