Zhang Yi-Fan, Zhang Zhang, Li Da, Jia Zhen, Wang Liang, Tan Tieniu
IEEE Trans Image Process. 2023;32:509-523. doi: 10.1109/TIP.2022.3229621. Epub 2022 Dec 30.
Generalizable person Re-Identification (ReID) aims to learn ready-to-use cross-domain representations for direct cross-data evaluation, which has attracted growing attention in the recent computer vision (CV) community. In this work, we construct a structural causal model (SCM) among identity labels, identity-specific factors (clothing/shoes color etc.), and domain-specific factors (background, viewpoints etc.). According to the causal analysis, we propose a novel Domain Invariant Representation Learning for generalizable person Re-Identification (DIR-ReID) framework. Specifically, we propose to disentangle the identity-specific and domain-specific factors into two independent feature spaces, based on which an effective backdoor adjustment approximate implementation is proposed for serving as a causal intervention towards the SCM. Extensive experiments have been conducted, showing that DIR-ReID outperforms state-of-the-art (SOTA) methods on large-scale domain generalization (DG) ReID benchmarks.
通用行人重识别（ReID）旨在学习可直接用于跨数据评估的即用型跨域表示，这在最近的计算机视觉（CV）社区中受到了越来越多的关注。在这项工作中，我们在身份标签、特定身份因素（衣服/鞋子颜色等）和特定领域因素（背景、视角等）之间构建了一个结构因果模型（SCM）。根据因果分析，我们提出了一种用于通用行人重识别的新颖的域不变表示学习（DIR-ReID）框架。具体来说，我们建议将特定身份和特定领域因素解缠到两个独立的特征空间中，并在此基础上提出一种有效的后门调整近似实现，作为对SCM的因果干预。我们进行了广泛的实验，结果表明DIR-ReID在大规模域泛化（DG）ReID基准上优于现有技术（SOTA）方法。
