RESEARCH ON SPORTS TEACHING AND TRAINING ACTION DETECTION BASED ON DEEP CONVOLUTION NEURAL NETWORK

Abstract

The aim of this study is to enhance the accuracy and efficacy of error action detection in sports coaching and training by employing a deep convolutional neural network (DCNN)-based approach. This method is designed to minimize the error rate associated with identifying incorrect movements during sports instruction and training sessions. After a thorough review of prior research findings, this study constructs a deep convolutional neural network (DCNN) focused on detecting errors in physical education teaching and training actions. The process begins with establishing feature extraction datasets, which are then fed into the network's input layer. The subsequent convolutional layers generate feature maps, and a normalization layer is integrated to refine the processing of physical education teaching and training samples. The error detection capability is achieved through iterative convolutional operations within the network. Experimental validation of this approach reveals an error rate of approximately 0.034%, indicating that the DCNN-based technique for physical education teaching and training action detection is highly precise in identifying training errors among athletes.