Unsupervised Region Attention Network for Person Re-Identification
As supervised person re-identification (Re-Id) requires massive labeled pedestrian data and it is very difficult to collect sufficient labeled data in reality, unsupervised Re-Id approaches attract much more attention than the former. Existing unsupervised person Re-Id models learn global features o...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2019-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8897584/ |
| _version_ | 1831669307725053952 |
|---|---|
| author | Chenrui Zhang Yangxu Wu Tao Lei |
| author_facet | Chenrui Zhang Yangxu Wu Tao Lei |
| author_sort | Chenrui Zhang |
| collection | DOAJ |
| description | As supervised person re-identification (Re-Id) requires massive labeled pedestrian data and it is very difficult to collect sufficient labeled data in reality, unsupervised Re-Id approaches attract much more attention than the former. Existing unsupervised person Re-Id models learn global features of pedestrian from whole images or several constant patches. These models ignore the difference of each region in the whole pedestrian images for feature representation, such as occluded and pose invariant regions, and thus reduce the robustness of models for cross-view feature learning. To solve these issues, we propose an Unsupervised Region Attention Network (URAN) that can learn the cross-view region attention features from the cropped pedestrian images, fixed by region importance weights on images. The proposed URAN designs a Pedestrian Region Biased Enhance (PRBE) loss to produce high attention weights for most important regions in pedestrian images. Furthermore, the URAN employs a first neighbor relation grouping algorithm and a First Neighbor Relation Constraint (FNRC) loss to provide the training direction of the unsupervised region attention network, such that the region attention features are discriminant enough for unsupervised person Re-Id task. In experiments, we consider two popular datasets, Market1501 and DukeMTMC-reID, as evaluation of PRBE and FNRC loss, and their balance parameter to demonstrate the effectiveness and efficiency of the proposed URAN, and the experimental results show that the URAN provides better performance than the-state-of-the-arts (higher than existing methods at least 1.1%). |
| first_indexed | 2024-12-19T22:56:52Z |
| format | Article |
| id | doaj.art-645a7d2ebf36495da3e843fc9bf9aeb8 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-19T22:56:52Z |
| publishDate | 2019-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-645a7d2ebf36495da3e843fc9bf9aeb82022-12-21T20:02:38ZengIEEEIEEE Access2169-35362019-01-01716552016553010.1109/ACCESS.2019.29532808897584Unsupervised Region Attention Network for Person Re-IdentificationChenrui Zhang0https://orcid.org/0000-0001-6788-0004Yangxu Wu1https://orcid.org/0000-0002-2116-6101Tao Lei2https://orcid.org/0000-0002-6952-377XSchool of Information and Communication Engineering, North University of China, Taiyuan, ChinaSchool of Information and Communication Engineering, North University of China, Taiyuan, ChinaSchool of Electronic Information and Artificial Intelligence, Shaanxi University of Science and Technology, Xi’an, ChinaAs supervised person re-identification (Re-Id) requires massive labeled pedestrian data and it is very difficult to collect sufficient labeled data in reality, unsupervised Re-Id approaches attract much more attention than the former. Existing unsupervised person Re-Id models learn global features of pedestrian from whole images or several constant patches. These models ignore the difference of each region in the whole pedestrian images for feature representation, such as occluded and pose invariant regions, and thus reduce the robustness of models for cross-view feature learning. To solve these issues, we propose an Unsupervised Region Attention Network (URAN) that can learn the cross-view region attention features from the cropped pedestrian images, fixed by region importance weights on images. The proposed URAN designs a Pedestrian Region Biased Enhance (PRBE) loss to produce high attention weights for most important regions in pedestrian images. Furthermore, the URAN employs a first neighbor relation grouping algorithm and a First Neighbor Relation Constraint (FNRC) loss to provide the training direction of the unsupervised region attention network, such that the region attention features are discriminant enough for unsupervised person Re-Id task. In experiments, we consider two popular datasets, Market1501 and DukeMTMC-reID, as evaluation of PRBE and FNRC loss, and their balance parameter to demonstrate the effectiveness and efficiency of the proposed URAN, and the experimental results show that the URAN provides better performance than the-state-of-the-arts (higher than existing methods at least 1.1%).https://ieeexplore.ieee.org/document/8897584/Unsupervised person re-identificationregion attentionfirst neighbor relationocclusionpose variant |
| spellingShingle | Chenrui Zhang Yangxu Wu Tao Lei Unsupervised Region Attention Network for Person Re-Identification IEEE Access Unsupervised person re-identification region attention first neighbor relation occlusion pose variant |
| title | Unsupervised Region Attention Network for Person Re-Identification |
| title_full | Unsupervised Region Attention Network for Person Re-Identification |
| title_fullStr | Unsupervised Region Attention Network for Person Re-Identification |
| title_full_unstemmed | Unsupervised Region Attention Network for Person Re-Identification |
| title_short | Unsupervised Region Attention Network for Person Re-Identification |
| title_sort | unsupervised region attention network for person re identification |
| topic | Unsupervised person re-identification region attention first neighbor relation occlusion pose variant |
| url | https://ieeexplore.ieee.org/document/8897584/ |
| work_keys_str_mv | AT chenruizhang unsupervisedregionattentionnetworkforpersonreidentification AT yangxuwu unsupervisedregionattentionnetworkforpersonreidentification AT taolei unsupervisedregionattentionnetworkforpersonreidentification |