Tennis video analysis
The business goal
The goal of the project is to build a platform for tennis analytics. During this project our main goal was to analyze the video of a tennis game in order to breakup games into shorter videos and highlight the most important sequences. It was required to remove those parts of the match where the players did not play (the players’ rest, the gap between the points, etc.); this allowed game statisticians to make further revisions of the game much faster because all “idle” periods of the game were removed and the total length was much shorter.
The logic of breaking up videos was developed based on the analysis of game events that were detected in the video, position, speed and posture of players, ball movement and location, and other parameters. We used the following Computer Vision and Machine Learning techniques: optic flow, background subtraction, HoG detector, pose detection and others.
The project resulted in building a successful Minimum Viable Product (MVP), generating funds at around $300,000 for the further enhancement of this platform and competing with such players as PlaySight. The future of the app is a platform for self-study and learning how to play tennis from the best players/coaches.