Computer-vision-based analysis of the temporal characteristics of straight punches in boxing

The application of computer vision technology in sports training is rapidly emerging. This study establishes a method for analysing the temporal characteristics of boxing straight punch movements based on skeleton point tracking technology, proposes a key frame identification algorithm and motion segmentation model, and constructs a core technical feature evaluation system. Through an anomaly detection model, real-time assessment and corrective guidance of motion standardisation are achieved, significantly improving training effectiveness and efficiency. The research findings provide new technical support for the modernisation of boxing training and have good practical application value.