Research map / Label-efficient and robust learning / Recognition and long-tail
Improving Deep Visual Representation for Person Re-identification by Global and Local Image-language Association
Dapeng Chen, Hongsheng Li, Xihui Liu, Yantao Shen, Jing Shao, Zejian Yuan, Xiaogang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add global discriminative association (GDA) (ablation) | ResNet-50 baseline · Market-1501 | mAP: 74.4 -> 80.0 | Improved |
| add global discriminative association (GDA) (ablation) | ResNet-50 baseline · CUHK-SYSU | mAP: 85.8 -> 90.2 | Improved |
| add local reconstructive association (LRA) (ablation) | ResNet-50 baseline · Market-1501 | mAP: 74.4 -> 79.6 | Improved |
| add local reconstructive association (LRA) (ablation) | ResNet-50 baseline · CUHK-SYSU | mAP: 85.8 -> 89.7 | Improved |
| add LRA to GDA (full method) (ablation) | basel.+GDA · Market-1501 | mAP: 80.0 -> 81.8 | Improved |
| add GDA to LRA (full method) (ablation) | basel.+LRA · Market-1501 | mAP: 79.6 -> 81.8 | Improved |
| use L_dis instead of ranking loss variant 2 (ablation) | basel.+rank2 · Market-1501 | mAP: 77.7 -> 80.0 | Improved |
| use L_dis instead of ranking loss variant 2 (ablation) | basel.+rank2 · CUHK-SYSU | mAP: 88.8 -> 90.2 | Improved |
| use more positive pairs (any image-text of same person) in ranking loss (ablation) | basel.+rank1 · Market-1501 | mAP: 75.5 -> 77.7 | Improved |
| add language ID loss L_T (lambda_T 0 -> 0.1) (ablation) | basel.+GDA · Market-1501 | mAP: 78.9 -> 80.0 | Improved |
| add language ID loss L_T (lambda_T 0 -> 0.1) (ablation) | basel.+GDA · CUHK-SYSU | mAP: 89.3 -> 90.2 | Improved |
| raise lambda_T from 0.1 to 1 (ablation) | basel.+GDA · Market-1501 | mAP: 80.0 -> 78.9 | Hurt |
Closest papers by meaning
- ChatReID: Open-ended Interactive Person Retrieval via Hierarchical Progressive Tuning for Vision Language Models (ICCV 2025 · similarity 0.88)
- Vision-Language Attribute Disentanglement and Reinforcement for Lifelong Person Re-Identification (CVPR 2026 · similarity 0.88)
- Deep Learning for Person Re-Identification: A Survey and Outlook (TPAMI 2021 · similarity 0.88)
- AXM-Net: Implicit Cross-Modal Feature Alignment for Person Re-identification (AAAI 2022 · similarity 0.88)
- Unifying Multi-Modal Uncertainty Modeling and Semantic Alignment for Text-to-Image Person Re-identification (AAAI 2024 · similarity 0.87)
- Noisy-Correspondence Learning for Text-to-Image Person Re-identification (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
