This paper explores the latest research advancements in domain generalization for person re-identifi-cation within the framework of deep learning,aiming to provide a reference for future research and applications.Firstly,it reviews the basic workflow of person re-identification and related work in cross-domain scenarios.Secondly,it provides a detailed analysis of research work in four areas concerning domain generalization issues:domain-invariant feature representation learning,multi-expert mixture methods,meta-learning methods,and deep graph matching methods.Finally,it prospects the future research directions for domain generalization in person re-identification.
关键词
行人重识别/域泛化/深度学习/计算机视觉
Key words
person re-identification/domain generalization/deep learning/computer vision
维普
万方数据