GAN-Based Data Augmentation and Pseudo-label Refinement for Unsupervised Domain Adaptation Person Re-identification

Person re-identification (re-ID) by using an unsupervised domain adaptation (UDA) approach has drawn considerable attention in contemporary security research. Thus, UDA person re-ID usually employs a model learned from a labeled source domain, adjusted by pseudo-labels, for an unlabeled target domain. However, this method still needs to overcome two main challenges: a significant gap between the source and target domains and the accuracy of pseudo-labels generated by a clustering algorithm. To address these problems, we propose a novel method to improve UDA person re-ID performance by combining GAN-based Data Augmentation and Unsupervised Pseudo-Label Editation methods for training on Target Domain, named DAUET. In particular, we first use a generative adversarial network (GAN) method to bridge the distribution of the source and target domains. Then we propose a supervised learning approach to maximize the benefits of the virtual dataset. Finally, we utilize a pseudo-label refinement to enhance the unsupervised learning process. Extensive experiments on two popular datasets, Market-1501 and DukeMTMC-reID, indicate that our DAUET method can substantially outperform the state-of-the-art performance of the UDA person re-ID.