Proceedings of the First International Conference on Science, Engineering and Technology Practices for Sustainable Development, ICSETPSD 2023, 17th-18th November 2023, Coimbatore, Tamilnadu, India

Research Article

A Comprehensive Statistical Evaluation System for Boxing Referees Based on Convolutional Neural Networks

Download32 downloads
  • @INPROCEEDINGS{10.4108/eai.17-11-2023.2342788,
        author={Chen  Chen},
        title={A Comprehensive Statistical Evaluation System for Boxing Referees Based on Convolutional Neural Networks},
        proceedings={Proceedings of the First International Conference on Science, Engineering and Technology Practices for Sustainable Development, ICSETPSD 2023, 17th-18th November 2023, Coimbatore, Tamilnadu, India},
        publisher={EAI},
        proceedings_a={ICSETPSD},
        year={2024},
        month={1},
        keywords={convolutional neural network boxing referee evaluation comprehensive statistics},
        doi={10.4108/eai.17-11-2023.2342788}
    }
    
  • Chen Chen
    Year: 2024
    A Comprehensive Statistical Evaluation System for Boxing Referees Based on Convolutional Neural Networks
    ICSETPSD
    EAI
    DOI: 10.4108/eai.17-11-2023.2342788
Chen Chen1,*
  • 1: Xi'an International Studies University, Sports Department, Xi'an, 710128, Shaanxi, China
*Contact email: 15529635333@163.com

Abstract

In boxing, as the referee makes decisions through punting, a referee evaluation system allows for adequate analysis of the vast amount of punting data. The aim of this paper is to study a comprehensive statistical assessment system for boxing referees based on convolutional neural networks. The concept of comprehensive statistical assessment of referees is proposed and a data warehouse for comprehensive statistical assessment of referees is established. A multi-dimensional data query analysis is realised. Development of a video retrieval module for boxing systems using video retrieval technology. LeNet-5 convolutional neural network-based recognition of athlete's punching situation is proposed. The different effects of the learning rate on the recognition effect are then investigated, and it is experimentally demonstrated that the best results are obtained at a learning rate of 0.001.