A Domain Adaptive Person Re-Identification Based on Dual Attention Mechanism and Camstyle Transfer
Chengyan Zhong,
Guanqiu Qi,
Neal Mazur,
Sarbani Banerjee,
Devanshi Malaviya,
Gang Hu
Affiliations
Chengyan Zhong
Key Laboratory of Industrial Internet of Things and Networked Control, Ministry of Education, College of Automation, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
Guanqiu Qi
Computer Information Systems Department, State University of New York at Buffalo State, Buffalo, NY 14222, USA
Neal Mazur
Computer Information Systems Department, State University of New York at Buffalo State, Buffalo, NY 14222, USA
Sarbani Banerjee
Computer Information Systems Department, State University of New York at Buffalo State, Buffalo, NY 14222, USA
Devanshi Malaviya
Computer Information Systems Department, State University of New York at Buffalo State, Buffalo, NY 14222, USA
Gang Hu
Computer Information Systems Department, State University of New York at Buffalo State, Buffalo, NY 14222, USA
Due to the variation in the image capturing process, the difference between source and target sets causes a challenge in unsupervised domain adaptation (UDA) on person re-identification (re-ID). Given a labeled source training set and an unlabeled target training set, this paper focuses on improving the generalization ability of the re-ID model on the target testing set. The proposed method enforces two properties at the same time: (1) camera invariance is achieved through the positive learning formed by unlabeled target images and their camera style transfer counterparts; and (2) the robustness of the backbone network feature extraction is improved, and the accuracy of feature extraction is enhanced by adding a position-channel dual attention mechanism. The proposed network model uses a classic dual-stream network. Comparative experimental results on three public benchmarks prove the superiority of the proposed method.
