JAC-Net: Joint learning with adaptive exploration and concise attention for unsupervised domain adaptive person re-identification

Existing unsupervised domain adaptive (UDA) methods of person re-identification (re-ID) often use clus-tering to generate and optimize pseudo-labels. However, the pseudo-labels generated in this way contain noise, which is gradually amplified during the iterative process, leading to a lower recognition accuracy than for supervised methods. This paper proposes Joint Learning with Adaptive Exploration and Concise Attention Network (JAC-Net), which uses two identical networks to optimize person re-ID in an unlabeled target domain. Based on the pseudo-labels generated by clustering, JAC-Net optimizes the training net-work by combining a joint learning network (JLN) with a concise attention module (CAM). Inspired by the teacher-student network, JLN uses two identical networks to share knowledge for network learning, and also applies adaptive exploration learning strategies to automatically assign weights to the two iden-tical networks and to balance the impact of the knowledge from the two networks. As a parameter-free attention module, a CAM is added to the feature map extracted in specific layers of ResNet50 without changing the high-order semantic features. Extensive experiments on the Market-1501, DukeMTMC-reID and MSMT17 datasets show that JAC-Net achieves well performance and reaches a similar level of supervised learning. (C) 2022 Elsevier B.V. All rights reserved.

Automated summary

This paper proposes a method called JAC-Net for person re-identification in an unlabeled target domain. It uses clustering to generate pseudo-labels and optimizes the training network using a joint learning network and a concise attention module. Experiments show that JAC-Net achieves good performance on multiple datasets and reaches a similar level to supervised learning methods.