Research map / Label-efficient and robust learning / Recognition and long-tail
Relation Network for Person Re-Identification
Read the paperCheck 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 |
|---|---|---|---|
| Use P2, P4, and P6 multi-scale features rather than P6 alone (ablation) | Relation Network · Market1501 | mAP: 88.0 -> 88.9 | Helped |
| Use one-vs.-rest relational features instead of independent part features (ablation) | Relation Network · Market1501 | mAP: 83.7 -> 86.7 | Helped |
| Use one-vs.-rest relational features instead of independent part features (ablation) | Relation Network · Market1501 | rank-1: 93.2 -> 94.2 | Helped |
| Use GCP instead of GMP for the global feature (ablation) | Relation Network with relation module · Market1501 | mAP: 86.7 -> 88.0 | Helped |
| Use GCP instead of GAP+GMP for the global feature (ablation) | Relation Network with relation module · Market1501 | mAP: 86.6 -> 88.0 | Helped |
| Use GMP rather than GAP for local and global pooling (ablation) | Person representation without relation module · Market1501 | mAP: 82.9 -> 83.7 | Helped |
| Use the part and rest features together rather than the rest feature alone (ablation) | One-vs.-rest relation module · Market1501 | mAP: 84.5 -> 86.7 | Helped |
| Use the part and rest features together rather than the rest feature alone (ablation) | One-vs.-rest relation module · Market1501 | rank-1: 93.4 -> 94.2 | Helped |
| Use GCP rather than GAP+GMP for the single global descriptor (ablation) | Global feature · Market1501 | mAP: 82.5 -> 84.6 | Helped |
| Use GCP rather than GAP+GMP for the single global descriptor (ablation) | Global feature · Market1501 | rank-1: 92.6 -> 93.4 | Helped |
| Use both local and global features instead of a global feature alone (ablation) | Person representation · Market1501 | mAP: 74.5 -> 82.9 | Helped |
| Use the relation module on a single local feature (ablation) | Part-level features · Market1501 | mAP: 29.6 -> 58.7; 46.4 -> 68.5; 55.3 -> 73.5; 51.8 -> 71.9; 49.7 -> 69.7; 35.5 -> 55.7 | Helped |
Closest papers by meaning
- Person30K: A Dual-Meta Generalization Network for Person Re-Identification (CVPR 2021 · similarity 0.91)
- Learning Deep Context-Aware Features Over Body and Latent Parts for Person Re-Identification (CVPR 2017 · similarity 0.90)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.89)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.89)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · similarity 0.88)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · 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.