Research map / Label-efficient and robust learning / Recognition and long-tail
Keypoint Promptable Re-Identification
Vladimir Somers, Alexandre Alahi, Christophe De Vleeschouwer
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 keypoint prompts to KPR (ablation) | KPR_IN · Occluded PoseTrack-ReID | Rank-1: 85.3 -> 92.3 | Improved |
| add keypoint prompts to KPR (ablation) | KPR_IN · Occluded-Duke | mAP: 63.4 -> 67.1 | Improved |
| add keypoint prompts to KPR (ablation) | KPR_SOL · Market-1501 | mAP: 93.0 -> 93.2 | No clear effect |
| add keypoint prompts to KPR (ablation) | KPR_SOL · Partial-ReID | Rank-1: 90.3 -> 90.7 | No clear effect |
| add keypoint prompts to KPR (ablation) | KPR_SOL · Occluded-ReID | Rank-1: 80.0 -> 84.8 | Improved |
| add part-based head (global to part-based embeddings) (ablation) | KPR baseline · Occluded PoseTrack-ReID | Rank-1: 84.8 -> 83.3 on Occ-PTrack; 61.8 -> 67.6 on Occluded-Duke | Mixed |
| add multi-scale features module (ablation) | KPR baseline with part-based head · Occluded-Duke | mAP: 54.2 -> 56.8 | Improved |
| add positive keypoint prompts (ablation) | KPR · Occluded PoseTrack-ReID | Rank-1: 84.8 -> 88.8 | Improved |
| add negative keypoint prompts (ablation) | KPR with positive prompts · Occluded PoseTrack-ReID | Rank-1: 88.8 -> 90.0 | Improved |
| add BIPO batch-wise inter-person occlusion augmentation with prompts (ablation) | KPR · Occluded PoseTrack-ReID | Rank-1: 90.0 -> 92.3 | Improved |
| use SOLIDER pre-trained weights instead of ImageNet (ablation) | KPR · Occluded PoseTrack-ReID | Rank-1: 92.3 -> 90.6 | Hurt |
| use SOLIDER pre-trained weights instead of ImageNet (ablation) | KPR · Occluded-Duke | mAP: 67.1 -> 75.1 | Improved |
Closest papers by meaning
- Occluded Person Re-Identification With Single-Scale Global Representations (ICCV 2021 · similarity 0.92)
- Dynamic Feature Pruning and Consolidation for Occluded Person Re-identification (AAAI 2024 · similarity 0.89)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.89)
- Multi-Prompts Learning with Cross-Modal Alignment for Attribute-Based Person Re-identification (AAAI 2024 · similarity 0.88)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · similarity 0.88)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · similarity 0.88)
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.
