摘要
行人搜索是计算机视觉领域中重要的研究方向之一,旨在从未剪裁的图像库中检测和识别人物。为深入了解行人搜索算法,总结与分析大量相关文献。首先根据网络结构的不同,将行人搜索算法分为两类:一类是传统的两步法,一类是基于端到端的一步法,并重点分析和介绍一步法的关键技术——特征学习和度量学习。然后,介绍了行人搜索领域的数据集和评价指标,比较与分析主流算法性能。实验结果表明,两步法虽然实现了很好的性能,但大多数的方法计算成本较高,且耗时较长;而一步法可以在更高效的学习框架中共同解决行人检测和行人重识别2个子任务,效果更好。最后,总结行人搜索算法,并展望了未来的发展方向。
Person search is one of the important research directions in the field of computer vision.Its research goal is to detect and identify characters in uncropped image libraries.In order to deeply understand the person search algorithms,a large number of related literature were summarized and analyzed.First of all,according to the network structure,the person search algorithms were divided into two categories:two-step methods and end-to-end one-step methods.The key technologies of the one-step methods,feature learning and measurement learning,were analyzed and introduced.The datasets and evaluation indicators in the field of person search were discussed,and the performance comparison and analysis of the mainstream algorithms were given.The experimental results show that,although the two-step methods have good performance,most of them have high calculation costs and take long time;the one-step methods can solve the two sub-tasks pedestrian detection and person re-identification,in a more efficient learning framework and achieve better results.Finally,the person search algorithms were summarized and their future development directions were prospected.
作者
王翠
邓淼磊
张德贤
李磊
杨晓艳
WANG Cui;DENG Miaolei;ZHANG Dexian;LI Lei;YANG Xiaoyan(College of Information Science and Engineering,Henan University of Technology,Zhengzhou Henan 450001,China;Henan International Joint Laboratory of Grain Information Processing,Zhengzhou Henan 450001,China)
出处
《计算机应用》
CSCD
北大核心
2024年第8期2544-2550,共7页
journal of Computer Applications
基金
国家自然科学基金资助项目(62276091)
河南省重大公益专项(201300311200)。
关键词
行人搜索
两步法
一步法
行人检测
行人重识别
person search
two-step method
one-step method
pedestrian detection
person Re-IDentification(ReID)
作者简介
王翠(1998-),女,河南商丘人,硕士研究生,主要研究方向:行人搜索、计算机视觉;通信作者:邓淼磊(1977-),男,河南南阳人,教授,博士,CCF会员,主要研究方向:信息安全、物联网,电子邮箱:dengmiaolei@haut.edu.cn;张德贤(1961-),男,河南郑州人,教授,博士,主要研究方向:模式识别、智能信息处理;李磊(1979-),男,河南郑州人,副教授,博士,CCF会员,主要研究方向:图像处理、计算机视觉、机器学习;杨晓艳(1998-),女,河南许昌人,硕士研究生,主要研究方向:人脸识别。
