基于多分支协作的行人重识别网络  

Multi-branch Cooperative Network for Person Re-identification

作　　者：张磊[1] 吴晓富[1] 张索非[2] 尹梓睿 ZHANG Lei;WU Xiaofu;ZHANG Suofei;YIN Zirui(College of Telecommunication and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003;School of Internet of Things,Nanjing University of Posts and Telecommunications,Nanjing 210003)

机构地区：[1]南京邮电大学通信与信息工程学院,南京210003 [2]南京邮电大学物联网学院,南京210003

出　　处：《模式识别与人工智能》2021年第9期853-862,共10页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金项目(No.61701252)资助。

摘　　要：设计多分支网络以提取分集特征已成为行人重识别领域的重要方向之一.由于单分支学习到的特征表达能力有限,所以文中提出基于多分支协作的行人重识别网络.在局部分支、全局分支、全局对比池化分支、关联分支这4个相互协作的分支上进行特征提取,获得强大的分集行人特征表达能力.文中网络可应用于不同的主干网络.实验中考虑OSNet、ResNet作为文中网络的主干网络进行验证.实验表明,文中网络在行人重识别数据集上均取得Start-of-the-art结果.Designing multi-branch networks to learn rich feature representation is one of the important directions in person re-identification(Re-ID).Aiming at the limited feature representation learned by a single branch,a multi-branch cooperative network for person Re-ID(BC-Net)is proposed.Powerful feature representation for person Re-ID is obtained by extracting features from four cooperative branches,local branch,global branch,relational branch and contrastive branch.The proposed network can be applied to different backbone networks.OSNet and ResNet are considered as the backbone of the proposed network for verification.Extensive experiments show that BC-Net achieves state-of-the-art performance on the popular Re-ID datasets.

关 键 词：行人重识别 特征表示 深度学习 多分支网络结构 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]