Research map / Label-efficient and robust learning / Recognition and long-tail
Re-Ranking Person Re-Identification With k-Reciprocal Encoding
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 k-reciprocal re-ranking to IDE (R) + KISSME | Market-1501 · Market-1501 | mAP: 49.05 -> 63.63 | Helped |
| Add k-reciprocal re-ranking to IDE (R) + XQDA | CUHK03 new protocol · CUHK03 labeled, new protocol | mAP: 29.6 -> 40.3 | Helped |
| Add k-reciprocal re-ranking to IDE (R) + XQDA | CUHK03 new protocol · CUHK03 detected, new protocol | mAP: 28.2 -> 37.4 | Helped |
| Add k-reciprocal re-ranking to IDE (R) + XQDA | MARS · MARS | mAP: 55.12 -> 68.45 | Helped |
| Add k-reciprocal re-ranking to IDE (C) | PRW · PRW | mAP: 25.09 -> 31.51 | Helped |
| Add k-reciprocal re-ranking to IDE (C) | CUHK03 single-shot · CUHK03 labeled, single-shot | rank-1: 57.0 -> 57.2 | No clear effect |
| Add k-reciprocal re-ranking to IDE (C) + XQDA | CUHK03 single-shot · CUHK03 detected, single-shot | mAP: 64.9 -> 64.7 | Hurt |
| Use a single-shot gallery with only one positive per identity | k-reciprocal re-ranking · CUHK03 single-shot | rank-1 | No clear effect |
| Use local query expansion with an appropriate k2 instead of k2=1 (ablation) | k-reciprocal re-ranking · Market-1501 validation | rank-1 | Helped |
| Set k2 too large in local query expansion (ablation) | k-reciprocal re-ranking · Market-1501 validation | rank-1 | Hurt |
| Increase k1 beyond its optimal range (ablation) | k-reciprocal re-ranking · Market-1501 validation | rank-1 | Hurt |
| Use Jaccard distance alone instead of original distance alone (ablation) | k-reciprocal re-ranking · Market-1501 validation | rank-1 | Helped |
Closest papers by meaning
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · similarity 0.91)
- A Pose-Sensitive Embedding for Person Re-Identification With Expanded Cross Neighborhood Re-Ranking (CVPR 2018 · similarity 0.89)
- CA-Jaccard: Camera-aware Jaccard Distance for Person Re-identification (CVPR 2024 · similarity 0.89)
- Re-Ranking via Metric Fusion for Object Retrieval and Person Re-Identification (CVPR 2019 · similarity 0.88)
- Semi-Supervised Bayesian Attribute Learning for Person Re-Identification (AAAI 2018 · similarity 0.86)
- Learning Continual Compatible Representation for Re-indexing Free Lifelong Person Re-identification (CVPR 2024 · similarity 0.85)
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.