Journal of Computer Applications ›› 2019, Vol. 39 ›› Issue (8): 2235-2241.DOI: 10.11772/j.issn.1001-9081.2019010084

• Artificial intelligence • Previous Articles     Next Articles

Aggressive behavior recognition based on human joint point data

CHEN Hao, XIAO Lixue, LI Guang, PAN Yuekai, XIA Yu   

  1. School of Computer Science & Technology, Xi'an University of Posts & Telecommunications, Xi'an Shaanxi 710121, China
  • Received:2019-01-15 Revised:2019-03-08 Online:2019-08-10 Published:2019-04-15
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61203311, 61105064), the Scientific Research Program Funded by Shaanxi Provincial Education Department of China (17JK0701), the Science Foundation of the Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing (XUPT-KLND (201804)), the Graduate Innovation Fund of Xi'an University of Posts & Telecommunications (CXJJ2017069).

基于人体关节点数据的攻击性行为识别

陈皓, 肖利雪, 李广, 潘跃凯, 夏雨   

  1. 西安邮电大学 计算机学院, 西安 710121
  • 通讯作者: 陈皓
  • 作者简介:陈皓(1978-),男,河北安新人,副教授,博士,CCF会员,主要研究方向:进化计算、工程优化;肖利雪(1992-),女,内蒙古赤峰人,硕士研究生,主要研究方向:计算机智能、数据挖掘;李广(1995-),男,陕西铜川人,硕士研究生,主要研究方向:计算机智能、数据挖掘;潘跃凯(1995-),男,山东聊城人,硕士研究生,主要研究方向:计算机智能、数据挖掘;夏雨(1996-),女,陕西咸阳人,硕士研究生,主要研究方向:计算机智能、数据挖掘。
  • 基金资助:
    国家自然科学基金资助项目(61203311,61105064);陕西省教育厅自然科学专项(17JK0701);陕西省网络数据分析与智能处理重点实验室开放课题基金资助项目(XUPT-KLND (201804));西安邮电大学创新基金资助项目(CXJJ2017069)。

Abstract: In order to solve the problem of human aggressive behavior recognition, an aggressive behavior recognition method based on human joint points was proposed. Firstly, OpenPose was used to obtain the human joint point data of a single frame image, and nearest neighbor frame feature weighting method and piecewise polynomial regression were used to realize the completion of missing values caused by body self-occlusion and environmental factors. Then, the dynamic "safe distance" threshold was defined for each human body. If the true distance between the two people was less than the threshold, the behavior feature vector was constructed, including the human barycenter displacement between frames, the angular velocity of human joint rotation and the minimum attack distance during interaction. Finally, the improved LightGBM (Light Gradient Boosting Machine) algorithm, namely w-LightGBM (weight LightGBM), was used to realize the classification and recognition of aggressive behaviors. The public dataset UT-interaction was used to verify the proposed method, and the accuracy reached 95.45%. The results show that this method can effectively identify the aggressive behaviors from various angles.

Key words: human joint point data, aggressive behavior recognition, missing value completion, attack distance

摘要: 针对人体攻击性行为识别问题,提出一种基于人体关节点数据的攻击性行为识别方法。首先,利用OpenPose获得单帧图像中的人体关节点数据,并通过最近邻帧特征加权法和分段多项式回归完成由人体自遮挡和环境因素所导致缺失值的补全;然后,对每个人体定义动态"安全距离"阈值,如果两人真实距离小于阈值,则构建行为特征矢量,其中包括帧间人体重心位移、人体关节旋转角角速度和发生交互时的最小攻击距离等;最后,提出改进的LightGBM算法w-LightGBM,并对攻击性行为进行识别。采用公共数据集UT-interaction对所提出的攻击性行为分类识别方法进行测试实验,准确率达到95.45%。实验结果表明,所提方法能够有效识别各种角度的攻击性行为。

关键词: 人体关节点数据, 攻击性行为识别, 缺失值补全, 攻击距离

CLC Number: