Construction of evaluation system of 24-style Tai Chi classic technique movements based on computer vision

This paper constructs a 24-type Tai chi classic technique action evaluation system based on computer vision, adopts the literature method and other research methods, constructs a video data set containing high, middle and low quality actions, builds an evaluation system, builds an evaluation model based on human joint points, uses CNN convolutional neural network to extract joint points, classify and score, and encapsulates the model to develop the corresponding system. It has functions such as recording data, reading and analyzing video. The system supports automatic evaluation, avoids the disadvantages of traditional teaching, and promotes teaching interaction and communication.