Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss
Despite the promising progress made in recent years, person re-identification (Re-ID) remains a challenging task due to the intra-class variations. Most of the current studies used the traditional Softmax loss for solutions, but its discriminative capability encounters a bottleneck. Therefore, how t...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9195852/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832540554138746880 |
|---|---|
| author | Guofa Li Lisha Huang Liangwen Tang Chunli Han Yaoyu Chen Heng Xie Shen Li Gang Xu |
| author_facet | Guofa Li Lisha Huang Liangwen Tang Chunli Han Yaoyu Chen Heng Xie Shen Li Gang Xu |
| author_sort | Guofa Li |
| collection | DOAJ |
| description | Despite the promising progress made in recent years, person re-identification (Re-ID) remains a challenging task due to the intra-class variations. Most of the current studies used the traditional Softmax loss for solutions, but its discriminative capability encounters a bottleneck. Therefore, how to improve person Re-ID performance is still a challenging task. To address this problem, we proposed a novel loss function, namely additive distance constraint with similar labels loss (ADCSLL). Specifically, we reformulated the Softmax loss by adding a distance constraint to the ground truth label, based on which similar labels were introduced to enhance the learned features to be much more stable and centralized. Experimental evaluations were conducted on two popular datasets (Market-1501 and DukeMTMC-reID) to examine the effectiveness of our proposed method. The results showed that our proposed ADCSLL was more discriminative than most of the other compared state-of-the-art methods. The rank-1 accuracy and the mAP on Market-1501 were 95.0% and 87.0%, respectively. The numbers were 88.6% and 77.2% on DukeMTMC-reID, respectively. |
| format | Article |
| id | doaj-art-b9f2f5d419d0471c8b170b9e9b958b43 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-b9f2f5d419d0471c8b170b9e9b958b432025-02-05T00:00:40ZengIEEEIEEE Access2169-35362020-01-01816811116812010.1109/ACCESS.2020.30239489195852Person Re-Identification Using Additive Distance Constraint With Similar Labels LossGuofa Li0https://orcid.org/0000-0002-7889-4695Lisha Huang1Liangwen Tang2Chunli Han3Yaoyu Chen4Heng Xie5Shen Li6https://orcid.org/0000-0002-7111-8861Gang Xu7College of Mechatronics and Control Engineering, Institute of Human Factors and Ergonomics, Shenzhen University, Shenzhen, ChinaCollege of Mechatronics and Control Engineering, Institute of Human Factors and Ergonomics, Shenzhen University, Shenzhen, ChinaShenzhen Suanzi Technology Ltd., Shenzhen, ChinaHangzhou Nicigo Technology Company Ltd., Hangzhou, ChinaCollege of Mechatronics and Control Engineering, Institute of Human Factors and Ergonomics, Shenzhen University, Shenzhen, ChinaCollege of Mechatronics and Control Engineering, Institute of Human Factors and Ergonomics, Shenzhen University, Shenzhen, ChinaDepartment of Civil and Environmental Engineering, University of Wisconsin–Madison, Madison, WI, USAShenzhen Key Laboratory of Urban Rail Transit, College of Urban Transportation and Logistics, Shenzhen Technology University, Shenzhen, ChinaDespite the promising progress made in recent years, person re-identification (Re-ID) remains a challenging task due to the intra-class variations. Most of the current studies used the traditional Softmax loss for solutions, but its discriminative capability encounters a bottleneck. Therefore, how to improve person Re-ID performance is still a challenging task. To address this problem, we proposed a novel loss function, namely additive distance constraint with similar labels loss (ADCSLL). Specifically, we reformulated the Softmax loss by adding a distance constraint to the ground truth label, based on which similar labels were introduced to enhance the learned features to be much more stable and centralized. Experimental evaluations were conducted on two popular datasets (Market-1501 and DukeMTMC-reID) to examine the effectiveness of our proposed method. The results showed that our proposed ADCSLL was more discriminative than most of the other compared state-of-the-art methods. The rank-1 accuracy and the mAP on Market-1501 were 95.0% and 87.0%, respectively. The numbers were 88.6% and 77.2% on DukeMTMC-reID, respectively.https://ieeexplore.ieee.org/document/9195852/Intelligent safety systemsperson re-identificationdeep learningsimilar labelsdistance constraint |
| spellingShingle | Guofa Li Lisha Huang Liangwen Tang Chunli Han Yaoyu Chen Heng Xie Shen Li Gang Xu Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss IEEE Access Intelligent safety systems person re-identification deep learning similar labels distance constraint |
| title | Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss |
| title_full | Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss |
| title_fullStr | Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss |
| title_full_unstemmed | Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss |
| title_short | Person Re-Identification Using Additive Distance Constraint With Similar Labels Loss |
| title_sort | person re identification using additive distance constraint with similar labels loss |
| topic | Intelligent safety systems person re-identification deep learning similar labels distance constraint |
| url | https://ieeexplore.ieee.org/document/9195852/ |
| work_keys_str_mv | AT guofali personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT lishahuang personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT liangwentang personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT chunlihan personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT yaoyuchen personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT hengxie personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT shenli personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss AT gangxu personreidentificationusingadditivedistanceconstraintwithsimilarlabelsloss |