A Novel Approach to Pedestrian Re-Identification in Low-Light and Zero-Shot Scenarios: Exploring Transposed Convolutional Reflectance Decoders
Abstract
:1. Introduction
- Our work narrows the performance gap in pedestrian re-identification under low-light conditions, a scenario that is underrepresented in current research.
- We improved the reflectance representation learning module originally applied in the zero-shot daytime domain adaptation network for object detection and proposed a transposed convolution reflectance decoder (TransConvRefDecoder) for pedestrian re-identification.
- We conducted extensive experiments on multiple datasets to validate the effectiveness of our proposed method, demonstrating significant improvements over baseline models.
2. Related Works
2.1. Pedestrian Re-Identification
2.2. Low-Light Image Enhancement
2.3. Low-Light Pedestrian Re-Identification
3. Methods
3.1. Overview
3.2. Pedestrian Re-Identification Network
3.2.1. Backbone
3.2.2. Reflectance Decoder
3.2.3. Retinex DecomNet
3.3. Training
4. Experiments
4.1. Settings
4.2. Results
4.3. Analysis
4.4. Ablation Study
4.4.1. Backbone
4.4.2. TransConvRefDecoder
4.4.3. Robustness
Dataset | Methods | R-1 | R-5 | R-10 | mAP |
---|---|---|---|---|---|
Market1501 | VGG16 | 68.6 | 81.1 | 84.9 | 17.0 |
VGG16 + ReflectanceDecoder | 68.9 | 81.6 | 85.5 | 20.5 | |
VGG16 + TransConvRefDecoder | 74.9 | 85.9 | 89.3 | 21.4 | |
VGG16 + TransConvRefDecoder + ABMLP | 75.5 | 86.9 | 89.8 | 22.1 | |
CUHK03 | VGG16 | 62.6 | 80.4 | 85.0 | 24.7 |
VGG16 + ReflectanceDecoder | 64.4 | 81.7 | 87.6 | 25.3 | |
VGG16 + TransConvRefDecoder | 67.2 | 84.4 | 88.6 | 26.2 | |
VGG16 + TransConvRefDecoder + ABMLP | 66.6 | 83.2 | 87.8 | 26.2 | |
MSMT17 | VGG16 | 40.4 | 54.8 | 60.6 | 7.6 |
VGG16 + ReflectanceDecoder | 41.9 | 55.8 | 61.6 | 8.1 | |
VGG16 + TransConvRefDecoder | 42.2 | 56.6 | 62.4 | 8.1 | |
VGG16 + TransConvRefDecoder + ABMLP | 42.4 | 57.0 | 61.5 | 8.2 |
5. Conclusions
5.1. Conclusions
5.2. Limitations and Future Work
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
References
- Gao, Z.; Gao, L.; Zhang, H.; Cheng, Z.; Hong, R.; Chen, S. DCR: A Unified Framework for Holistic/Partial Person ReID. IEEE Trans. Multimed. 2021, 23, 3332–3345. [Google Scholar] [CrossRef]
- Wu, D.; Ye, M.; Lin, G.; Gao, X.; Shen, J. Person Re-Identification by Context-Aware Part Attention and Multi-Head Collaborative Learning. IEEE Trans. Inf. Forensics Secur. 2021, 17, 115–126. [Google Scholar] [CrossRef]
- Ye, M.; Li, H.; Du, B.; Shen, J.; Shao, L.; Hoi, S.C.H. Collaborative Refining for Person Re-Identification with Label Noise. IEEE Trans. Image Process. 2021, 31, 379–391. [Google Scholar] [CrossRef] [PubMed]
- Zeng, Z.; Wang, Z.; Wang, Z.; Chuang, Y.Y.; Satoh, S. Illumination-Adaptive Person Re-identification. IEEE Trans. Multimed. 2019, 22, 3064–3074. [Google Scholar] [CrossRef]
- Luo, H.; Jiang, W.; Gu, Y.; Liu, F.; Liao, X.; Lai, S.; Gu, J. A Strong Baseline and Batch Normalization Neck for Deep Person Re-Identification. IEEE Trans. Multimed. 2019, 22, 2597–2609. [Google Scholar] [CrossRef]
- Zhang, S.; Zhang, Q.; Wei, X.; Wang, P.; Jiao, B.; Zhang, Y. Person Re-Identification in Aerial Imagery. IEEE Trans. Multimed. 2019, 23, 281–291. [Google Scholar] [CrossRef]
- Jia, M.; Cheng, X.; Lu, S.; Zhang, J. Learning Disentangled Representation Implicitly Via Transformer for Occluded Person Re-Identification. IEEE Trans. Multimed. 2021, 25, 1294–1305. [Google Scholar] [CrossRef]
- Wei, C.; Wang, W.; Yang, W.; Liu, J. Deep Retinex Decomposition for Low-Light Enhancement. arXiv 2018, arXiv:1808.04560. [Google Scholar]
- Wu, W.B.; Weng, J.; Zhang, P.; Wang, X.; Yang, W.; Jiang, J. URetinex-Net: Retinex-based Deep Unfolding Network for Low-light Image Enhancement. In Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA, 18–24 June 2022; pp. 5891–5900. [Google Scholar]
- Cui, Z.; Qi, G.J.; Gu, L.; You, S.; Zhang, Z.; Harada, T. Multitask AET with Orthogonal Tangent Regularity for Dark Object Detection. In Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 10–17 October 2021; pp. 2533–2542. [Google Scholar]
- Wang, W.; Yang, W.; Liu, J. HLA-Face: Joint High-Low Adaptation for Low Light Face Detection. In Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 20–25 June 2021; pp. 16190–16199. [Google Scholar]
- Lengyel, A.; Garg, S.; Milford, M.; van Gemert, J.C. Zero-Shot Day-Night Domain Adaptation with a Physics Prior. In Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 10–17 October 2021; pp. 4379–4389. [Google Scholar]
- Luo, R.; Wang, W.; Yang, W.; Liu, J. Similarity Min-Max: Zero-Shot Day-Night Domain Adaptation. In Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV), Paris, France, 1–6 October 2023; pp. 8070–8080. [Google Scholar]
- Du, Z.; Shi, M.; Deng, J. Boosting Object Detection with Zero-Shot Day-Night Domain Adaptation. arXiv 2023, arXiv:2312.01220. [Google Scholar]
- Zheng, Z.; Zheng, L.; Yang, Y. A Discriminatively Learned CNN Embedding for Person Reidentification. Acm Trans. Multimed. Comput. Commun. Appl. (Tomm) 2016, 14, 1–20. [Google Scholar] [CrossRef]
- Wu, L.; Shen, C.; van den Hengel, A. PersonNet: Person Re-identification with Deep Convolutional Neural Networks. arXiv 2016, arXiv:1601.07255. [Google Scholar]
- Sun, Y.; Zheng, L.; Yang, Y.; Tian, Q.; Wang, S. Beyond Part Models: Person Retrieval with Refined Part Pooling. In Proceedings of the European Conference on Computer Vision, Amsterdam, The Netherlands, 8–14 October 2017. [Google Scholar]
- Li, W.; Zhu, X.; Gong, S. Harmonious Attention Network for Person Re-identification. In Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 18–23 June 2018; pp. 2285–2294. [Google Scholar]
- Qian, X.; Fu, Y.; Xiang, T.; Wang, W.; Qiu, J.; Wu, Y.; Jiang, Y.G.; Xue, X. Pose-Normalized Image Generation for Person Re-identification. In Proceedings of the European Conference on Computer Vision, Amsterdam, The Netherlands, 8–14 October 2017. [Google Scholar]
- Feng, J.; Wu, A.; Zheng, W.S. Shape-Erased Feature Learning for Visible-Infrared Person Re-Identification. In Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, BC, Canada, 17–24 June 2023; pp. 22752–22761. [Google Scholar]
- Zhang, Y.; Yan, Y.; Li, J.; Wang, H. MRCN: A Novel Modality Restitution and Compensation Network for Visible-Infrared Person Re-identification. In Proceedings of the 37th AAAI Conference on Artificial Intelligence, Washington, DC, USA, 7–14 February 2023. [Google Scholar]
- Lu, A.; Zhang, Z.; Huang, Y.; Zhang, Y.; Li, C.; Tang, J.; Wang, L. Illumination Distillation Framework for Nighttime Person Re-Identification and a New Benchmark. IEEE Trans. Multimed. 2023, 26, 406–419. [Google Scholar] [CrossRef]
- Simonyan, K.; Zisserman, A. Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv 2014, arXiv:1409.1556. [Google Scholar]
- Land, E.H. The retinex theory of color vision. Sci. Am. 1977, 237, 108–128. [Google Scholar] [CrossRef]
- Rukundo, O.; Cao, H. Nearest Neighbor Value Interpolation. arXiv 2012, arXiv:1211.1768. [Google Scholar]
- Ma, H.; Lei, S.; Celik, T.; Li, H.C. FER-YOLO-Mamba: Facial Expression Detection and Classification Based on Selective State Space. arXiv 2024, arXiv:2405.01828. [Google Scholar]
- Schroff, F.; Kalenichenko, D.; Philbin, J. FaceNet: A unified embedding for face recognition and clustering. In Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, USA, 7–12 June 2015; pp. 815–823. [Google Scholar]
- Mao, A.; Mohri, M.; Zhong, Y. Cross-Entropy Loss Functions: Theoretical Analysis and Applications. arXiv 2023, arXiv:2304.07288. [Google Scholar]
- Willmott, C.J.; Matsuura, K. Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Clim. Res. 2005, 30, 79–82. [Google Scholar] [CrossRef]
- Wang, Z.; Bovik, A.C.; Sheikh, H.R.; Simoncelli, E.P. Image Quality Assessment: From Error Measurement to Structural Similarity. IEEE Trans. Image Process. 2004, 13, 600–613. [Google Scholar] [CrossRef]
- Zheng, L.; Shen, L.; Tian, L.; Wang, S.; Bu, J.; Tian, Q. Person Re-identification Meets Image Search. arXiv 2015, arXiv:1502.02171. [Google Scholar]
- Li, W.; Zhao, R.; Xiao, T.; Wang, X. DeepReID: Deep Filter Pairing Neural Network for Person Re-identification. In Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 23–28 June 2014; pp. 152–159. [Google Scholar]
- Wei, L.; Zhang, S.; Gao, W.; Tian, Q. Person Transfer GAN to Bridge Domain Gap for Person Re-identification. In Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 18–23 June 2018; pp. 79–88. [Google Scholar]
- Deng, J.; Dong, W.; Socher, R.; Li, L.J.; Li, K.; Fei-Fei, L. ImageNet: A large-scale hierarchical image database. In Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition, Miami, FL, USA, 20–25 June 2009; pp. 248–255. [Google Scholar]
- Robbins, H.E. A Stochastic Approximation Method. Ann. Math. Stat. 1951, 22, 400–407. [Google Scholar] [CrossRef]
- Chen, T.; Ding, S.; Xie, J.; Yuan, Y.; Chen, W.; Yang, Y.; Ren, Z.; Wang, Z. ABD-Net: Attentive but Diverse Person Re-Identification. In Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Republic of Korea, 27 October–2 November 2019; pp. 8350–8360. [Google Scholar]
- Ye, M.; Shen, J.; Lin, G.; Xiang, T.; Shao, L.; Hoi, S.C.H. Deep Learning for Person Re-Identification: A Survey and Outlook. IEEE Trans. Pattern Anal. Mach. Intell. 2020, 44, 2872–2893. [Google Scholar] [CrossRef] [PubMed]
- He, S.; Luo, H.; Wang, P.; Wang, F.; Li, H.; Jiang, W. TransReID: Transformer-based Object Re-Identification. In Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 10–17 October 2021; pp. 14993–15002. [Google Scholar]
Decoder | Params (K) | GFLOPs | Time (ms) |
---|---|---|---|
ReflectanceDecoder | 38.7 | 2.9 | 2.6 |
TransConvRefDecoder | 38.7 | 2.5 | 1.0 |
TransConvRefDecoder + ABMLP | 39.2 | 2.5 | 1.2 |
Datasets | IDs | Train | Query | Gallery |
---|---|---|---|---|
Market1501 | 1501 | 123,936 | 3368 | 19,732 |
CUHK03 | 1467 | 7368 | 1400 | 5328 |
MSMT17 | 4101 | 32,621 | 11,659 | 82,161 |
Dataset | Methods | R-1 | R-5 | R-10 | mAP |
---|---|---|---|---|---|
Market1501 | BoT | 51.1 | 69.8 | 76.1 | 26.4 |
ABD-Net | 46.2 | 65.7 | 73.6 | 24.1 | |
AGW | 58.9 | 75.9 | 81.4 | 31.5 | |
TransReID | 50.8 | 71.8 | 77.9 | 26.7 | |
IDF | 55.1 | 73.9 | 80.6 | 31.0 | |
TransConvRefDecoder | 74.9 | 85.9 | 89.3 | 21.4 | |
TransConvRefDecoder+ABMLP | 75.5 | 86.9 | 89.8 | 22.1 | |
CUHK03 | BoT | 16.2 | 36.9 | 46.4 | 16.1 |
ABD-Net | 14.7 | 30.8 | 41.7 | 14.2 | |
AGW | 11.9 | 26.9 | 37.4 | 12.5 | |
TransReID | 6.0 | 15.8 | 23.6 | 6.8 | |
IDF | 20.9 | 40.7 | 51.0 | 17.0 | |
TransConvRefDecoder | 67.2 | 84.4 | 88.6 | 26.2 | |
TransConvRefDecoder+ABMLP | 66.6 | 83.2 | 87.8 | 26.2 | |
MSMT17 | BoT | 22.2 | 36.2 | 42.7 | 8.1 |
ABD-Net | 36.2 | 56.1 | 66.9 | 10.2 | |
AGW | 24.2 | 35.4 | 41.0 | 7.8 | |
TransReID | 22.6 | 35.4 | 42.5 | 10.7 | |
IDF | 20.8 | 34.6 | 41.0 | 8.5 | |
TransConvRefDecoder | 42.2 | 56.6 | 62.4 | 8.1 | |
TransConvRefDecoder+ABMLP | 42.4 | 57.0 | 61.5 | 8.2 |
Dataset | Backbone | R-1 | R-5 | R-10 | mAP |
---|---|---|---|---|---|
Market1501 | VGG16 + TransConvRefDecoder | 74.9 | 85.9 | 89.3 | 21.4 |
ResNet18 + TransConvRefDecoder | 54.7 | 70.3 | 75.9 | 12.6 | |
EfficientNet + TransConvRefDecoder | 38.9 | 60.5 | 69.1 | 9.8 | |
CUHK03 | VGG16 + TransConvRefDecoder | 67.2 | 84.4 | 88.6 | 26.2 |
ResNet18 + TransConvRefDecoder | 51.9 | 74.1 | 81.1 | 21.0 | |
EfficientNet + TransConvRefDecoder | 45.6 | 66.6 | 78.0 | 16.6 | |
MSMT17 | VGG16 + TransConvRefDecoder | 42.2 | 56.6 | 62.4 | 8.1 |
ResNet18 + TransConvRefDecoder | 39.4 | 51.8 | 60.9 | 6.8 | |
EfficientNet + TransConvRefDecoder | 37.0 | 45.6 | 55.8 | 5.7 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Li, Z.; Xiong, J. A Novel Approach to Pedestrian Re-Identification in Low-Light and Zero-Shot Scenarios: Exploring Transposed Convolutional Reflectance Decoders. Electronics 2024, 13, 4069. https://doi.org/10.3390/electronics13204069
Li Z, Xiong J. A Novel Approach to Pedestrian Re-Identification in Low-Light and Zero-Shot Scenarios: Exploring Transposed Convolutional Reflectance Decoders. Electronics. 2024; 13(20):4069. https://doi.org/10.3390/electronics13204069
Chicago/Turabian StyleLi, Zhenghao, and Jiping Xiong. 2024. "A Novel Approach to Pedestrian Re-Identification in Low-Light and Zero-Shot Scenarios: Exploring Transposed Convolutional Reflectance Decoders" Electronics 13, no. 20: 4069. https://doi.org/10.3390/electronics13204069
APA StyleLi, Z., & Xiong, J. (2024). A Novel Approach to Pedestrian Re-Identification in Low-Light and Zero-Shot Scenarios: Exploring Transposed Convolutional Reflectance Decoders. Electronics, 13(20), 4069. https://doi.org/10.3390/electronics13204069