A feature enhancement loss for person re-identification
The goal of person re-identification (ReID) is to recognize the same person across cameras. Classification loss is one of the most widely used objective functions in person ReID tasks based on deep learning. However, the features, which are learned with the classification loss, are not sufficiently...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2023-12-01
|
| Series: | Systems Science & Control Engineering |
| Subjects: | |
| Online Access: | https://www.tandfonline.com/doi/10.1080/21642583.2023.2220482 |
| _version_ | 1797447512100962304 |
|---|---|
| author | Yao Peng Yining Lin Huajian Ni Hua Gao Chenchen Hu |
| author_facet | Yao Peng Yining Lin Huajian Ni Hua Gao Chenchen Hu |
| author_sort | Yao Peng |
| collection | DOAJ |
| description | The goal of person re-identification (ReID) is to recognize the same person across cameras. Classification loss is one of the most widely used objective functions in person ReID tasks based on deep learning. However, the features, which are learned with the classification loss, are not sufficiently discriminative enough when they are close to the origin. In this study, we propose a feature enhancement loss to move features of person images away from the origin. During training, our proposed method adds a regularization item to avoid the feature vector near the origin point. Our method was evaluated on two benchmark person ReID benchmark datasets, Market1501 and DukeMTMC-reID. Results show that the proposed method outperforms the state-of-the-art method by 0.9% and 1.2% on rank-1 accuracy and mean average precision (mAP) index on Market-1501, 1.0% and 1.4% on rank-1 accuracy and mAP index on DukeMTMC-reID. This means that when learning features with a classification loss, making the features far away from the origin point is meaningful. |
| first_indexed | 2024-03-09T13:57:04Z |
| format | Article |
| id | doaj.art-728935c2d74c40fab19f08d4a9e1b20b |
| institution | Directory Open Access Journal |
| issn | 2164-2583 |
| language | English |
| last_indexed | 2024-03-09T13:57:04Z |
| publishDate | 2023-12-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | Systems Science & Control Engineering |
| spelling | doaj.art-728935c2d74c40fab19f08d4a9e1b20b2023-11-30T12:45:32ZengTaylor & Francis GroupSystems Science & Control Engineering2164-25832023-12-0111110.1080/21642583.2023.2220482A feature enhancement loss for person re-identificationYao Peng0Yining Lin1Huajian Ni2Hua Gao3Chenchen Hu4Shanghai SUPREMIND Technology Co., Ltd., Shanghai, People's Republic of ChinaShanghai SUPREMIND Technology Co., Ltd., Shanghai, People's Republic of ChinaShanghai SUPREMIND Technology Co., Ltd., Shanghai, People's Republic of ChinaCollege of Computer Science and Technology, Zhejiang University of Technology, Hangzhou, People's Republic of ChinaCollege of Computer Science and Technology, Zhejiang University of Technology, Hangzhou, People's Republic of ChinaThe goal of person re-identification (ReID) is to recognize the same person across cameras. Classification loss is one of the most widely used objective functions in person ReID tasks based on deep learning. However, the features, which are learned with the classification loss, are not sufficiently discriminative enough when they are close to the origin. In this study, we propose a feature enhancement loss to move features of person images away from the origin. During training, our proposed method adds a regularization item to avoid the feature vector near the origin point. Our method was evaluated on two benchmark person ReID benchmark datasets, Market1501 and DukeMTMC-reID. Results show that the proposed method outperforms the state-of-the-art method by 0.9% and 1.2% on rank-1 accuracy and mean average precision (mAP) index on Market-1501, 1.0% and 1.4% on rank-1 accuracy and mAP index on DukeMTMC-reID. This means that when learning features with a classification loss, making the features far away from the origin point is meaningful.https://www.tandfonline.com/doi/10.1080/21642583.2023.2220482Classification lossperson re-identificationdeep learning |
| spellingShingle | Yao Peng Yining Lin Huajian Ni Hua Gao Chenchen Hu A feature enhancement loss for person re-identification Systems Science & Control Engineering Classification loss person re-identification deep learning |
| title | A feature enhancement loss for person re-identification |
| title_full | A feature enhancement loss for person re-identification |
| title_fullStr | A feature enhancement loss for person re-identification |
| title_full_unstemmed | A feature enhancement loss for person re-identification |
| title_short | A feature enhancement loss for person re-identification |
| title_sort | feature enhancement loss for person re identification |
| topic | Classification loss person re-identification deep learning |
| url | https://www.tandfonline.com/doi/10.1080/21642583.2023.2220482 |
| work_keys_str_mv | AT yaopeng afeatureenhancementlossforpersonreidentification AT yininglin afeatureenhancementlossforpersonreidentification AT huajianni afeatureenhancementlossforpersonreidentification AT huagao afeatureenhancementlossforpersonreidentification AT chenchenhu afeatureenhancementlossforpersonreidentification AT yaopeng featureenhancementlossforpersonreidentification AT yininglin featureenhancementlossforpersonreidentification AT huajianni featureenhancementlossforpersonreidentification AT huagao featureenhancementlossforpersonreidentification AT chenchenhu featureenhancementlossforpersonreidentification |