摘要
基于OpenPose算法的研究与改进,从目标检测、特征提取及分类、行为识别算法等理论问题研究入手,进行系统方案设计,开发针对自动扶梯乘降人员摔倒等异常行为的智能监控及预警系统。通过智能算法对乘降人员摔倒等异常行为的分析,及时对自动扶梯的运行状态进行预警报警,从而提高人员乘坐自动扶梯的安全性。
Based on the improvement of the Open Pose algorithm,with the target detection,feature extraction and classification,and behavior recognition algorithms,we designed a system to develop an intelligent monitoring and early warning for abnormal behaviors such as escalator riders falling.Through intelligent algorithm analysis of abnormal behaviors,such as falls of passengers,timely warning and alarming of the operation status of escalators are carried out,which improved the safety of personnel riding on escalators.
作者
张鑫鑫
杨超宇
杨成
Zhang Xinxin;Yang Chaoyu;Yang Cheng(School of Economics and Management,Anhui University of Science&Technology,Huainan,Anhui 232000,China)
出处
《黑龙江工业学院学报（综合版）》
2020年第10期49-55,共7页
Journal of Heilongjiang University of Technology(Comprehensive Edition)
基金
国家自然科学基金项目“多源传感器环境下基于异构特征信息融合的行为识别”(项目编号:61873004)
安徽理工大学研究生创新基金项目“基于计算机视觉的自动扶梯人员乘降和物品运输安全监控系统研究”(项目编号:2019CX2073)。
关键词
OpenPose
特征分类
行为识别
智能监控
Open Pose
feature classification
behavior recognition
Intelligent monitoring
