Abstract:Recent years have witnessed outstanding success in supervised domain adaptive person re-identification (ReID). However, the model often suffers serious performance drops when transferring to another domain in real-world applications. To address the domain gap situations, many unsupervised domain adaptive (UDA) methods have been proposed to adapt the model trained on the source domain to a target domain. Such methods are typically based on clustering algorithms to generate pseudo labels. Noisy labels, which often exist due to the instability of clustering algorithms, will substantially affect the performance of UDA methods. In this study, we focused on intermediate domains that can be regarded as a bridge that connects source and target domains. We added a domainness factor in the loss function of SPGAN that can decide the style of the image generated by the GAN model. We obtained a series of intermediate domains by changing the value of the domainness factor. Pseudo labels are more reliable because intermediate domains are closer to the source domain compared with the target domain. We then fine-tuned the model pre-trained with source data on these intermediate domains. The fine-tuning process was conducted repeatedly because intermediate domains are composed of more than one dataset. Finally, the model fine-tuned on intermediate domains was adapted to the target domain. The model can easily adapt to changes in image style as we gradually transfer the model to the target domain along the bridge consisting of several intermediate domains. To the best of our knowledge, we are the first to apply intermediate domains to UDA problems. We evaluated our method on Market1501, DukeMTMC-reID and MSMT17 datasets. Experimental results proved that our method brings a significant improvement and achieves a state-of-the-art performance.

Memory Bank for Unsupervised Domain Adaptation Person Retrieval

Unsupervised Domain Adaptation With Class-Aware Memory Alignment

Black-box Unsupervised Domain Adaptation with Bi-directional Atkinson-Shiffrin Memory

Memory-Assisted Sub-Prototype Mining for Universal Domain Adaptation

Memorizing Comprehensively to Learn Adaptively: Unsupervised Cross-Domain Person Re-ID with Multi-level Memory

Unsupervised Multi-Source Domain Adaptation for Person Re-Identification.

Adaptive Memorization with Group Labels for Unsupervised Person Re-identification

Unsupervised Domain Adaptation via Style-Aware Self-intermediate Domain

Retrievable Domain-Sensitive Feature Memory for Multi-Domain Recommendation

Learning deep part-aware embedding for person retrieval

A semi-supervised cross-modal memory bank for cross-modal retrieval

Unsupervised Domain Adaptive Person Re-Identification via Human Learning Imitation

Lifelong Unsupervised Domain Adaptive Person Re-identification with Coordinated Anti-forgetting and Adaptation

DREAMT: Diversity Enlarged Mutual Teaching for Unsupervised Domain Adaptive Person Re-Identification.

Attention-disentangled Re-Id Network for Unsupervised Domain Adaptive Person Re-Identification

Unsupervised Person Re-Identification Based on Removal of Camera Bias and Dynamic Updating of the Memory Bank

Semantic Driven Attention Network with Attribute Learning for Unsupervised Person Re-Identification

Learning to Adapt Invariance in Memory for Person Re-Identification

Unsupervised domain adaptation semantic segmentation of high-resolution remote sensing imagery with invariant domain-level prototype memory

CMFT: Contrastive Memory Feature Transfer for Non-shared-and-Imbalanced Unsupervised Domain Adaption

Unsupervised Domain Adaptive Person Re-Identification Via Intermediate Domains