基于姿态识别的武术套路评分系统的设计  

Design of Martial Arts Routine Scoring System Based on Posture Recognition

作　　者：夏永强 周元玲[1] XIA Yongqiang;ZHOU Yuanling(Hunan University of Humanities,Science and Technology,Loudi,Hunan Province,417000 China)

机构地区：[1]湖南人文科技学院,湖南娄底417000

出　　处：《当代体育科技》2023年第6期121-124,共4页Contemporary Sports Technology

基　　金：2021年湖南省大学生创新训练计划项目(S20110553032);湖南人文科技学院大学生创新创业训练计划资助项目(校教通[2021]56号-38)。

摘　　要：现有的武术套路竞赛评分主要采用人工评分的方式,主观性较强。该文提出将人体姿态识别计算应用到武术套路竞赛评分中。通过Mediapipe为基础库和视觉预处理的OpenCV2库进行人体姿态检测,通过人体姿态算法计算相关关节点的角度,实现武术套路竞赛的量化评估。通过实验表明,该系统具有较强的计算和分析能力,能够通过关节角度等反映运动者的动作质量,减少主观性评分。The existing scoring of martial arts routine competitions mainly adopts the manual scoring method,which is highly subjective.This paper proposes to apply the calculation of human posture recognition to the scoring of martial arts routine competitions for quantitative scoring.The OpenCV2 library based on Mediapipe and visual preprocessing is used to detect the human posture,and the angle of the relevant points is calculated by the human posture algorithm to realize the quantitative evaluation of martial arts routine competitions.Experiments show that the system has strong computation and analysis ability,can reflect the movement quality of athletes through joint angles,and reduce subjective score.

关 键 词：姿态识别 武术套路竞赛 量化评分 OPENCV 

分 类 号：G80[文化科学—运动人体科学]