During competitive ball games, people usually use the information system to collect skill-related data according to players’ performance. Based on the analyzed outcome, athletes can improve their training methods and the simulation of their tactical skills. Traditionally we record the process of competitions and details in games through viewing the videos of plays over and over again, and mark the types of balls and how the player moves by manual work. Not only is it cumbersome and labor-intensive, it could also lead to false marking or missing important information. Thanks to the development of deep learning technology in the past few years, we are able to analyze the videos of competitions by using deep learning technology and video visual technology. With the combination of technology and sport, we are capable of acquiring data of the balls’ trajectory, the athletes’ movement and the postures of their skeleton. Additionally, it enhances athletes’ performance by providing paths for coaches and athletes to follow. Our research adopts TrackNet, which can identify and mark the high-speed flying balls in videos of competitions. After that we use OpenPose to detect the spot of player and obtain the player’s postures of their skeleton. With these methods we can record the types of balls the player used and analyze the speed and the exact landing spots of each shot during the games, in order to find out the nimbleness of the athlete, their habitual moves in games and so on. These data can not only provide coaches and athletes references for training, but is also important material for analyzing tactical skills in big data analysis. Athletes usually react each shot by their instinct during competitions. Now, by combining visual demonstration of the situation of competitions and development of assistive device, we can construct tactical training systems for both mental and physical base on the games’ big data database. Our research combines sport and science. Not only did we turn video records into digital data, we also convert big data analysis into concrete intelligence. In the future, it will without a doubt become a winning weapon in sport competitions.
Demo Video: