摘要
现今社会中的视频监控网络已日渐成为人们生活中不可缺少的安防体系,行人再识别技术正是智能视频监控的关键。为使读者对行人再识别问题有一个深入、整体的了解,本文首先回顾行人再识别的研究以及发展历程;然后针对当前研究热点,分别从基于特征表示、度量学习以及深度学习网络的角度将经典算法进行分类总结;其次介绍常用行人数据集、性能评价指标以及经典数据集的实验结果对比;最后对行人再识别今后的发展趋势作进一步展望。
The video surveillance equipment which is widely applied in today′s society has been indispensable security system,the key task in intelligent video surveillance is person re-identification (person re-id) exactly.The work of this paper will enable the readers to have an in-depth and overall understanding the person re-id problem.Firstly,the development process of person re-id is reviewed.Then the existing methods that based on feature representation,distance metric learning and deep learning network are summarized.Secondly,some comparison of experimental results on several commonly-used datasets,and performance evaluation standards,and common benchmark datasets are also introduced.Finally,the future develop direction of person re-id is discussed.
作者
胡正平
张敏姣
李淑芳
孙德刚
HU Zhengping;ZHANG Minjiao;LI Shufang;SUN Degang(School of Information Science and Engineering,Yanshan University,Qinhuangdao,Hebei 066004 China;Hebei Key Laboratory of Information Transmission and Signal Processing,Qinhuangdao,Hebei 066004,China;School of Electronic Information and Engineering,Shandong Huayu University of Technology,Dezhou,Shandong 253000,China)
出处
《燕山大学学报》
CAS
北大核心
2019年第5期377-393,共17页
Journal of Yanshan University
基金
国家自然科学基金资助项目(61071199)
河北省自然科学基金资助项目(F2016203422)
关键词
行人再识别
视频监控
距离度量
深度学习
person re-identification
video surveillance
metric learning
deep learning
