Unsupervised Pedestrian Re-Id: Attention-Driven Framework with Clustering Optimization

34 Pages Posted: 18 Aug 2023

See all articles by Xuan Wang

Xuan Wang

Yantai University

Zhaojie Sun

Yantai University

Abdellah Chehri

Royal Military College of Canada

Gwanggil Jeon

Incheon National University

Yongchao Song

Yantai University

Abstract

Unsupervised pedestrian re-identification (re-ID) is still a challenging task. Although convolutional neural networks (CNN) have had great success in pedestrian re-ID, they face the challenge of dealing with pose changes, occlusions, and lighting changes. To effectively tackle these challenges, it is imperative to prioritize the implementation of efficient sampling strategies. We propose an attention-driven framework with clustering optimization (AFC) to address the above issues. First, we introduce a new attention mechanism that enhances multi-scale spatial attention and reduces the number of trainable parameters. Then, we employed a straightforward and effective method of group sampling. In addition, we apply a clustering consensus approach to estimate pseudo-label similarity in continuous training and use temporal propagation and ensembles to improve pseudo-labels. Extensive experiments on Market-1501, DukeMTMC-reID, and MSMT17 have shown that our approach achieves significant performance gains in unsupervised pedestrian re-ID tasks, providing valuable insights for further research in this area.

Keywords: pattern recognition, Unsupervised Pedestrian re-identification, attention mechanism, Instance Learning, Group Sampling Method, Improve Pseudo-Labels.

Suggested Citation

Wang, Xuan and Sun, Zhaojie and Chehri, Abdellah and Jeon, Gwanggil and Song, Yongchao, Unsupervised Pedestrian Re-Id: Attention-Driven Framework with Clustering Optimization. Available at SSRN: https://ssrn.com/abstract=4545317 or http://dx.doi.org/10.2139/ssrn.4545317

Xuan Wang

Yantai University ( email )

32, Qingquan RD
Laishan District
Yantai, 264005
China

Zhaojie Sun

Yantai University ( email )

32, Qingquan RD
Laishan District
Yantai, 264005
China

Abdellah Chehri (Contact Author)

Royal Military College of Canada ( email )

Kingston
Canada

Gwanggil Jeon

Incheon National University ( email )

119 Academy-ro
Yeonsu-gu
Incheon
Korea, Republic of (South Korea)

Yongchao Song

Yantai University ( email )

32, Qingquan RD
Laishan District
Yantai, 264005
China

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
20
Abstract Views
94
PlumX Metrics