Journal of Computer Applications ›› 2021, Vol. 41 ›› Issue (5): 1458-1464.DOI: 10.11772/j.issn.1001-9081.2020071113

Special Issue: 多媒体计算与计算机仿真

• Virtual reality and multimedia computing • Previous Articles     Next Articles

Fitness action recognition method based on human skeleton feature encoding

GUO Tianxiao1, HU Qingrui1, LI Jianwei2, SHEN Yanfei2   

  1. 1. School of Sport Science, Beijing Sport University, Beijing 100084, China;
    2. School of Sports Engineering, Beijing Sport University, Beijing 100084, China
  • Received:2020-07-30 Revised:2020-09-25 Online:2021-05-10 Published:2020-10-19
  • Supported by:
    This work is partially supported by the National Key Research and Development Program of China (2018YFC2000600), the Fundamental Research Funds for Central Universities (2020056, 2020010).

基于人体骨架特征编码的健身动作识别方法

郭天晓1, 胡庆锐1, 李建伟2, 沈燕飞2   

  1. 1. 北京体育大学 运动人体科学学院, 北京 100084;
    2. 北京体育大学 体育工程学院, 北京 100084
  • 通讯作者: 李建伟
  • 作者简介:郭天晓(1996-),男,山西大同人,硕士研究生,主要研究方向:智能体育、体育视频分析;胡庆锐(1996-),男,安徽滁州人,硕士研究生,主要研究方向:智能体育、体育视频分析;李建伟(1987-),女,甘肃兰州人,讲师,博士,主要研究方向:SLAM、计算机视觉、智能体育;沈燕飞(1976-),男,江苏靖江人,教授,博士,主要研究方向:人工智能、智能视频分析、体育大数据。
  • 基金资助:
    国家重点研发计划项目(2018YFC2000600);中央高校基本科研业务费专项资金资助项目(校2020056,校2020010)。

Abstract: Fitness action recognition is the core of the intelligent fitness system. In order to improve the accuracy and speed of fitness action recognition algorithm, and reduce the influence of the global displacement of fitness actions on the recognition results, a fitness action recognition method based on human skeleton feature encoding was proposed which included three steps:firstly, the simplified human skeleton model was constructed, and the information of skeleton model's joint point coordinates was extracted through the human pose estimation technology; secondly, the action feature region was extracted by using the human central projection method in order to eliminate the influence of the global displacement on action recognition; finally, the feature region was encoded as the feature vector and input to a multi-classifier to realize the action recognition, at the same time the length of the feature vector was optimized for improving the recognition rate and speed. Experiment results showed that the proposed method achieved the recognition rate of 97.24% on the self-built fitness dataset with 28 types of fitness actions, which verified the effectiveness of this method to recognize different types of fitness actions; on the public KTH and Weizmann datasets, the recognition rates of the proposed method were 91.67% and 90% respectively, higher than those of other similar methods.

Key words: computer vision, action recognition, intelligent fitness, skeleton information, pose estimation

摘要: 健身动作识别是智能健身系统的核心环节。为了提高健身动作识别算法的精度和速度,并减少健身动作中人体整体位移对识别结果的影响,提出了一种基于人体骨架特征编码的健身动作识别方法。该方法包括三个步骤:首先,构建精简的人体骨架模型,并利用人体姿态估计技术提取骨架模型中各关节点的坐标信息;其次,利用人体中心投影法提取动作特征区域以消除人体整体位移对动作识别的影响;最后,将特征区域编码作为特征向量并输入多分类器进行动作识别,同时通过优化特征向量长度使识别率和速度达到最优。实验结果表明,本方法在包含28种动作的自建健身数据集上的动作识别率为97.24%,证明该方法能够有效识别各类健身动作;在公开的KTH和Weizmann数据集上,所提方法的动作识别率分别为91.67%和90%,优于其他同类型方法。

关键词: 计算机视觉, 动作识别, 智能健身, 骨架信息, 姿态估计

CLC Number: