Learning Clothes-irrelevant Cues for Clothes-Changing Person Re-identification
Jingyi Mu (Nanjing University of Science and Technology),* Yong Li (Nanjing University of Science and Technology), Jun Li (Nanjing University of Science and Technology), Jian Yang (Nanjing University of Science and Technology)
The 33rd British Machine Vision Conference
Abstract
Person re-identification (re-ID), aiming to match a target person in a series of cross-camera images, is a challenging problem when people change their clothes. Essentially, different clothes are used to learn distinctive features usually, resulting in a failed identification. To mitigate this issue, we propose a Clothes-Relevant information Erasure (CRE) module to drive the model to adaptively learn clothes-irrelevant cues by utilizing the person's semantic information to eliminate the clothes-relevant features. Furthermore, we introduce a Body Shape-Guided Attention (BSGA) module so that the model can learn richer and more discriminative features. Compared to the state-of-the-art baselines, the experimental results on three benchmark datasets show the effectiveness and superiority of our method in the clothes-changing re-ID task.Video
Citation
@inproceedings{Mu_2022_BMVC,
author = {Jingyi Mu and Yong Li and Jun Li and Jian Yang},
title = {Learning Clothes-irrelevant Cues for Clothes-Changing Person Re-identification},
booktitle = {33rd British Machine Vision Conference 2022, {BMVC} 2022, London, UK, November 21-24, 2022},
publisher = {{BMVA} Press},
year = {2022},
url = {https://bmvc2022.mpi-inf.mpg.de/0337.pdf}
}
Copyright © 2022
The British Machine Vision Conference is organised by