基于动作识别算法的健美操难度自动评分系统设计
DOI:
作者:
作者单位:

作者简介:

通讯作者:

基金项目:

安徽省体育社会科学研究项目(ASS2016203)?


Design of Automatic Scoring System of Aerobic Difficulty Based on Action Recognition Algorithm
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    为了提高健美操评分的精度ꎬ减少人为主观性评价的影响ꎬ设计基于动作识别算法的健美操难度自动评分系统ꎮ 系 统数据采集层利用体感传感器 Kinect 采集健美操运动员体感信息ꎬ输出健美操运动员动作图像ꎻ将所获取健美操动作图像使 用串口通信协议传输至数据处理层ꎻ数据处理层使用基于最大相关-最小冗余的动作识别算法ꎬ识别健美操动作后传输至应 用层ꎻ应用层按照健美操难度评分标准自动评价健美操难度ꎮ 实验结果显示:所设计系统识别健美操动作时ꎬ识别结果的均 方误差、平均绝对误差较小ꎬ识别精度较高ꎻ对 10 位运动员健美操动作难度系数评分值与实际难度系数的差值较小ꎬ评分结果 精度较高ꎬ可以为健美操竞技评分提供依据ꎮ

    Abstract:

    In order to improve the accuracy of aerobics scoring and reduce the influence of subjective evaluationꎬ an auto ̄ matic scoring system of aerobics difficulty based on action recognition algorithm is designed. The data acquisition layer of the system collects the aerobic athletes′ somatosensory information and outputs the athletes′ action images through Kinectꎻ the obtained aerobics action images are transmitted to the data processing layer with serial communication protocolꎻ the da ̄ ta processing layer uses the action recognition algorithm based on maximum correlation and minimum redundancy to recog ̄ nize the aerobics actions and then transmit them to the application layerꎻ the application layer automatically scores the aer ̄ obic difficulty according to the aerobic difficulty criterion. The experimental results show that: when the designed system i ̄ dentifies aerobic movementsꎬ the mean square error and the mean absolute error of the recognition results are relatively smallꎬ and the recognition accuracy is highꎻ the difference between the difficulty coefficient scoring results and the actual difficulty coefficients of 10 athletes′ aerobic movements is smallꎬ and the scoring accuracy is high. Thereforeꎬ it can pro ̄ vide reference for aerobic competition scoring.

    参考文献
    相似文献
    引证文献
引用本文

马倩倩ꎬ贺 莉.基于动作识别算法的健美操难度自动评分系统设计[J].西昌学院学报(自然科学版),2021,35(2):106-110.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2021-07-27