English

Adaptive Domain-Specific Normalization for Generalizable Person Re-Identification

Computer Vision and Pattern Recognition 2021-05-12 v2

Abstract

Although existing person re-identification (Re-ID) methods have shown impressive accuracy, most of them usually suffer from poor generalization on unseen target domain. Thus, generalizable person Re-ID has recently drawn increasing attention, which trains a model on source domains that generalizes well on unseen target domain without model updating. In this work, we propose a novel adaptive domain-specific normalization approach (AdsNorm) for generalizable person Re-ID. It describes unseen target domain as a combination of the known source ones, and explicitly learns domain-specific representation with target distribution to improve the model's generalization by a meta-learning pipeline. Specifically, AdsNorm utilizes batch normalization layers to collect individual source domains' characteristics, and maps source domains into a shared latent space by using these characteristics, where the domain relevance is measured by a distance function of different domain-specific normalization statistics and features. At the testing stage, AdsNorm projects images from unseen target domain into the same latent space, and adaptively integrates the domain-specific features carrying the source distributions by domain relevance for learning more generalizable aggregated representation on unseen target domain. Considering that target domain is unavailable during training, a meta-learning algorithm combined with a customized relation loss is proposed to optimize an effective and efficient ensemble model. Extensive experiments demonstrate that AdsNorm outperforms the state-of-the-art methods. The code is available at: https://github.com/hzphzp/AdsNorm.

Keywords

Cite

@article{arxiv.2105.03042,
  title  = {Adaptive Domain-Specific Normalization for Generalizable Person Re-Identification},
  author = {Jiawei Liu and Zhipeng Huang and Kecheng Zheng and Dong Liu and Xiaoyan Sun and Zheng-Jun Zha},
  journal= {arXiv preprint arXiv:2105.03042},
  year   = {2021}
}

Comments

Withdraw this paper for internal review. Since we were not familiar with the use of arXiv, our initial manuscript was uploaded by mistake and we found many inappropriate and unmodified parts of it (such as the experimental results in Table 2,3, the Equation 13). I am sorry to say that this work still needs to be further completed and we do not intend to use it for publication

R2 v1 2026-06-24T01:51:51.044Z