Research map / Label-efficient and robust learning / Recognition and long-tail
ANFI: Rethinking Neighbor Feature Interaction in Person Re-ID
Xulin Li, Yan Lu, Bin Liu, Jiaze Li, Qinhong Yang, Tao Gong, Qi Chu, Nenghai Yu
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 affinity relations over single-image features (ablation) | ANFI baseline · Market1501 Standard | mAP: 90.3 -> 94.2 | Improved |
| add affinity relations over single-image features (ablation) | ANFI baseline · Market1501 1-shot | mAP: 93.5 -> 92.5 | Hurt |
| add affinity relations over single-image features (ablation) | ANFI baseline · Market1501 Hard Query | mAP: 65.5 -> 63.1 | Hurt |
| add discrepancy relations to affinity relations (ablation) | ANFI · Market1501 1-shot | mAP: 92.5 -> 93.5 | Improved |
| add discrepancy relations to affinity relations (ablation) | ANFI · Market1501 Hard Query | mAP: 63.1 -> 65.3 | Improved |
| add discrepancy relations to affinity relations (ablation) | ANFI · Market1501 Standard | mAP: 94.2 -> 94.8 | Improved |
| add noise simulation in training (ablation) | ANFI · Market1501 Standard | mAP: 94.8 -> 95.1 | Improved |
| add relationship regularization Lrr (ablation) | ANFI · Market1501 Standard | mAP: 95.1 -> 95.7 | Improved |
| add relationship regularization Lrr (ablation) | ANFI · Market1501 1-shot | mAP: 93.8 -> 94.4 | Improved |
| add relationship regularization Lrr (ablation) | ANFI · Market1501 Hard Query | mAP: 65.9 -> 66.1 | Improved |
| replace learned adaptive weight w_i with test-time oracle nnr (ablation) | ANFI · Market1501 Standard and Hard Query | mAP: -0.1 Standard, -0.3 Hard Query | Hurt |
| use ANFI instead of affinity-based re-ranking and neighbor methods | ANFI · Market1501 Standard | delta mAP: +5.4 vs +4.9 CA-Jacc., +4.8 K-recip. | Improved |
Closest papers by meaning
- NFormer: Robust Person Re-Identification With Neighbor Transformer (CVPR 2022 · similarity 0.85)
- Self-Similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-Identification (ICCV 2019 · similarity 0.84)
- Relation-Aware Global Attention for Person Re-Identification (CVPR 2020 · similarity 0.83)
- Rewarded Semi-Supervised Re-Identification on Identities Rarely Crossing Camera Views (TPAMI 2023 · similarity 0.83)
- ChatReID: Open-ended Interactive Person Retrieval via Hierarchical Progressive Tuning for Vision Language Models (ICCV 2025 · similarity 0.83)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · similarity 0.82)
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.
