System and method for understanding sports events based on machine learning

Abstract

The invention discloses a sports event understanding system based on machine learning, which relates to the technical field of sports event information systems. The system includes a data acquisition unit and a data analysis unit, and the data acquisition unit is used to acquire video information of the game by using a camera device , and transmit the video information to the data analysis unit; the data analysis unit is used to receive the video information, detect, identify and track the targets in the arena, and learn the tracking data of different targets through machine learning to understand the game events and obtain the game understanding information. The video information of the game is obtained through the data acquisition unit, and the data analysis unit analyzes and processes the received video information to detect, identify and track the targets in the arena, learn the tracking data of different targets through machine learning, understand the game events, and accurately understand the entire game .

Description

technical field [0001] The invention relates to the technical field of sports event information systems, in particular to a system and method for understanding sports events based on machine learning. Background technique [0002] At this stage, understanding sports events mainly relies on three methods: first, relying on manual labor; second, collecting data through equipment and analyzing the data to understand the game; third, using computer data analysis to assist manual methods. [0003] The first method is mainly realized by manually performing game statistics and manually watching games or videos. For artificial game data statistics, such as in ball games, several statisticians in each game watch the game from multiple perspectives, and are responsible for observing the teams on both sides. For specific game events such as scoring, assists, fouls, free throws, etc. Manual recording on mobile phone. The main disadvantages of manual statistics mainly include: 1. Poor ...

AI Technical Summary

Problems solved by technology

3D cameras have many problems in practical applications. For example, the pixels of pictures collected by time-of-flight cameras are low, and Kinect cameras can only be used within a short distance.

However, it is very difficult to reconstruct 3D from 2D information. The existing algorithm is difficult to guarantee its accuracy in the real game environment. In addition, the blocking between targets in the game makes it more difficult.

Images