Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Person Re-Identification via Multi-Label Classification
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 MPLP multi-label prediction rather than train with single-class pseudo labels (ablation) | MMCL ReID training · Market-1501 | Rank-1: 49.0 -> 80.3 | Helped |
| Add MPLP multi-label prediction rather than train with single-class pseudo labels (ablation) | MMCL ReID training · DukeMTMC-reID | Rank-1: 42.0 -> 65.2 | Helped |
| Add CamStyle data augmentation (ablation) | MMCL + MPLP · Market-1501 | Rank-1: 66.6 -> 80.3 | Helped |
| Use similarity-threshold filtering instead of fixed-K KNN (ablation) | MPLP label prediction · Market-1501 | Rank-1: 73.3 -> 77.6 | Helped |
| Add cycle consistency to similarity-score label filtering (ablation) | MPLP label prediction · Market-1501 | Rank-1: 77.6 -> 80.3 | Helped |
| Use MMCL instead of cross-entropy with MPLP labels (ablation) | ReID training · Market-1501 | Rank-1: 65.0 -> 80.3 | Helped |
| Use MMCL instead of cross-entropy with single-class labels (ablation) | ReID training · DukeMTMC-reID | Rank-1: 31.5 -> 42.0 | Helped |
| Set MMCL coefficient δ to 5 rather than 1 (ablation) | MMCL · Market-1501 | Rank-1: 59.3 -> 80.3 | Helped |
| Increase δ beyond 5 (ablation) | MMCL · Market-1501 | Rank-1: 80.3 -> 78.1 | Hurt |
| Use all negative classes rather than mine hard negatives (ablation) | MMCL · Market-1501 and DukeMTMC-reID | Rank-1 | Hurt |
| Use too small a hard-negative mining ratio (ablation) | MMCL · Market-1501 and DukeMTMC-reID | Rank-1 | Hurt |
| Lower MPLP similarity threshold into [0.3,0.5] (ablation) | MPLP · Market-1501 and DukeMTMC-reID | Rank-1 | Hurt |
Closest papers by meaning
- Unsupervised Person Re-Identification by Soft Multilabel Learning (CVPR 2019 · similarity 0.91)
- Part-Based Pseudo Label Refinement for Unsupervised Person Re-Identification (CVPR 2022 · similarity 0.90)
- Group-aware Label Transfer for Domain Adaptive Person Re-identification (CVPR 2021 · similarity 0.90)
- Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification (ICLR 2020 · similarity 0.89)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · similarity 0.87)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · 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.