Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification
In recent years, person re-identification based on deep learning approaches has made great progress and achieved good results. However, many of the latest network design methods, which usually deploy ResNet or SENet as the backbone network, were originally designed for classification tasks. Since th...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9217451/ |
| _version_ | 1818734554183106560 |
|---|---|
| author | Zongjing Cao Hyo Jong Lee |
| author_facet | Zongjing Cao Hyo Jong Lee |
| author_sort | Zongjing Cao |
| collection | DOAJ |
| description | In recent years, person re-identification based on deep learning approaches has made great progress and achieved good results. However, many of the latest network design methods, which usually deploy ResNet or SENet as the backbone network, were originally designed for classification tasks. Since the person re-identification task is essentially different from the classification task, the structure of the backbone network should be modified accordingly. In this paper, we propose a retrieval network based on a multi-scale backbone architecture, which is specifically suitable for the person re-identification task. By constructing hierarchical residual-like connections within a single residual block, the model learns multi-scale discriminative features of pedestrian images. Unlike many state-of-the-art methods that use complex network structures and concatenate multi-branch features, our proposed retrieval network is implemented using only global features, simple triplet loss, and softmax with cross-entropy loss. The results of extensive experiments show that the proposed network has stronger fine-grained pedestrian representation ability, leading to performance gains for person re-identification tasks. Our proposed network achieves a rank-1 accuracy of 96.03% on the Market-1501 and 92.11% on DukeMTMC-reID datasets while only using global features. |
| first_indexed | 2024-12-18T00:07:12Z |
| format | Article |
| id | doaj.art-3d3f4eff296640169fec6a808c8c2824 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-18T00:07:12Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-3d3f4eff296640169fec6a808c8c28242022-12-21T21:27:46ZengIEEEIEEE Access2169-35362020-01-01818464418465510.1109/ACCESS.2020.30295949217451Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-IdentificationZongjing Cao0https://orcid.org/0000-0001-9715-0619Hyo Jong Lee1https://orcid.org/0000-0003-2581-5268Department of Computer Science and Engineering, Jeonbuk National University, Jeonju, South KoreaDepartment of Computer Science and Engineering, Jeonbuk National University, Jeonju, South KoreaIn recent years, person re-identification based on deep learning approaches has made great progress and achieved good results. However, many of the latest network design methods, which usually deploy ResNet or SENet as the backbone network, were originally designed for classification tasks. Since the person re-identification task is essentially different from the classification task, the structure of the backbone network should be modified accordingly. In this paper, we propose a retrieval network based on a multi-scale backbone architecture, which is specifically suitable for the person re-identification task. By constructing hierarchical residual-like connections within a single residual block, the model learns multi-scale discriminative features of pedestrian images. Unlike many state-of-the-art methods that use complex network structures and concatenate multi-branch features, our proposed retrieval network is implemented using only global features, simple triplet loss, and softmax with cross-entropy loss. The results of extensive experiments show that the proposed network has stronger fine-grained pedestrian representation ability, leading to performance gains for person re-identification tasks. Our proposed network achieves a rank-1 accuracy of 96.03% on the Market-1501 and 92.11% on DukeMTMC-reID datasets while only using global features.https://ieeexplore.ieee.org/document/9217451/Person re-identificationconvolutional neural networksdeep learning |
| spellingShingle | Zongjing Cao Hyo Jong Lee Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification IEEE Access Person re-identification convolutional neural networks deep learning |
| title | Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification |
| title_full | Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification |
| title_fullStr | Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification |
| title_full_unstemmed | Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification |
| title_short | Learning Multi-Scale Features and Batch-Normalized Global Features for Person Re-Identification |
| title_sort | learning multi scale features and batch normalized global features for person re identification |
| topic | Person re-identification convolutional neural networks deep learning |
| url | https://ieeexplore.ieee.org/document/9217451/ |
| work_keys_str_mv | AT zongjingcao learningmultiscalefeaturesandbatchnormalizedglobalfeaturesforpersonreidentification AT hyojonglee learningmultiscalefeaturesandbatchnormalizedglobalfeaturesforpersonreidentification |