Research on Athlete Behavior Recognition Technology in Sports Teaching Video Based on Deep Neural Network.

Xianpin Zhao,Akshi Kumar

doi:10.1155/2022/7260894

Xianpin Zhao, Akshi Kumar

Open Access

https://doi.org/10.1155/2022/7260894

Copy DOI

Abstract

In recent years, due to the simple design idea and good recognition effect, deep learning method has attracted more and more researchers' attention in computer vision tasks. Aiming at the problem of athlete behavior recognition in mass sports teaching video, this paper takes depth video as the research object and cuts the frame sequence as the input of depth neural network model, inspired by the successful application of depth neural network based on two-dimensional convolution in image detection and recognition. A depth neural network based on three-dimensional convolution is constructed to automatically learn the temporal and spatial characteristics of athletes' behavior. The training results on UTKinect-Action3D and MSR-Action3D public datasets show that the algorithm can correctly detect athletes' behaviors and actions and show stronger recognition ability to the algorithm compared with the images without clipping frames, which effectively improves the recognition effect of physical education teaching videos.

Highlights

With the development of society, more and more sports teaching videos have entered people’s daily life. e analysis of PE teaching video can more effectively improve the teaching effect
Object detection and human pose estimation for sports video are the basis of sports video analysis and understanding. e existing target detection and human pose estimation technologies have achieved good performance in the general scene detection task based on pictures [1], but there are few algorithms and data for target detection in sports video scenes
For a new data field, accurate athlete detection and pose estimation are important links in sports video analysis. e existing human target detection and pose estimation algorithms have achieved good performance in the general human body detection task, but they will detect athletes and spectators at the same time in the physical education teaching video, so they cannot further distinguish athletes’ targets, which will interfere with the subsequent video analysis

Summary

Introduction

With the development of society, more and more sports teaching videos have entered people’s daily life. e analysis of PE teaching video can more effectively improve the teaching effect. E existing target detection and human pose estimation technologies have achieved good performance in the general scene detection task based on pictures [1], but there are few algorithms and data for target detection in sports video scenes. E existing human target detection and pose estimation algorithms have achieved good performance in the general human body detection task, but they will detect athletes and spectators at the same time in the physical education teaching video, so they cannot further distinguish athletes’ targets, which will interfere with the subsequent video analysis. Taking the deep learning method as the key technology, VGG16 convolutional neural network is used to extract the highly abstract features of video frames, and an athlete behavior recognition algorithm in sports teaching video is designed and implemented

Related Work

Athlete Behavior Recognition Model Based on Deep Neural Networks

Experiment and Analysis

Findings

Method