Trajectory Tracking Method for Table Tennis from Broadcast Video

Abstract

Currently, we are developing a system that automatically acquires strategies of table tennis match from broadcast video. In this paper, we introduce three modules which are motion tracking system, strategy acquisition system, strategy display system, and especially discuss motion tracking system. During the match, an algorithm removes noise, and estimates the ball and player position. The ball trajectory is estimated by preprocessing with the white blog extraction process and RGB extraction process, and then the player’s skeleton position is estimated by CenterNet of the deep learning (DNN). The position of the ball hidden at the body is estimated using the Kalman fllter and the bicube interpolation method. In addition, the ball trajectory and rally trajectory are automatically converted into two-dimensional coordinates. Finally, the system makes the strategy visible to coaches. We discuss a future image of this study.