Identity-aware Feature Decoupling Learning for Clothing-change Person Re-identification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Haoxuan, Li, Bo, Niu, Guanglin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917889109393408
author Xu, Haoxuan
Li, Bo
Niu, Guanglin
author_facet Xu, Haoxuan
Li, Bo
Niu, Guanglin
contents Clothing-change person re-identification (CC Re-ID) has attracted increasing attention in recent years due to its application prospect. Most existing works struggle to adequately extract the ID-related information from the original RGB images. In this paper, we propose an Identity-aware Feature Decoupling (IFD) learning framework to mine identity-related features. Particularly, IFD exploits a dual stream architecture that consists of a main stream and an attention stream. The attention stream takes the clothing-masked images as inputs and derives the identity attention weights for effectively transferring the spatial knowledge to the main stream and highlighting the regions with abundant identity-related information. To eliminate the semantic gap between the inputs of two streams, we propose a clothing bias diminishing module specific to the main stream to regularize the features of clothing-relevant regions. Extensive experimental results demonstrate that our framework outperforms other baseline models on several widely-used CC Re-ID datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2501_05851
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Identity-aware Feature Decoupling Learning for Clothing-change Person Re-identification
Xu, Haoxuan
Li, Bo
Niu, Guanglin
Computer Vision and Pattern Recognition
Clothing-change person re-identification (CC Re-ID) has attracted increasing attention in recent years due to its application prospect. Most existing works struggle to adequately extract the ID-related information from the original RGB images. In this paper, we propose an Identity-aware Feature Decoupling (IFD) learning framework to mine identity-related features. Particularly, IFD exploits a dual stream architecture that consists of a main stream and an attention stream. The attention stream takes the clothing-masked images as inputs and derives the identity attention weights for effectively transferring the spatial knowledge to the main stream and highlighting the regions with abundant identity-related information. To eliminate the semantic gap between the inputs of two streams, we propose a clothing bias diminishing module specific to the main stream to regularize the features of clothing-relevant regions. Extensive experimental results demonstrate that our framework outperforms other baseline models on several widely-used CC Re-ID datasets.
title Identity-aware Feature Decoupling Learning for Clothing-change Person Re-identification
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2501.05851