Research map / Label-efficient and robust learning / Recognition and long-tail
MVP Matching: A Maximum-Value Perfect Matching for Mining Hard Samples, With Application to Person Re-Identification
Han Sun, Zhiyuan Chen, Shiyang Yan, Lin Xu
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 re-ranking at test time | MVP Loss · Market-1501 | mAP: 80.5% -> 90.9% | Improved |
| add re-ranking at test time | MVP Loss · Duke-MTMC | mAP: 70.0% -> 83.9% | Improved |
| add re-ranking at test time | MVP Loss · CUHK03-Detected | Top-1: 91.8% -> 96.4% | Improved |
| add re-ranking at test time | MVP Loss · CUHK03-Labeled | Top-1: 93.7% -> 97.7% | Improved |
| replace batch-hard triplet loss with MVP matching loss | ResNet-50 metric learning · Market-1501 | mAP: 76.9% -> 80.5% | Improved |
| replace batch-wise OT loss with MVP matching loss | ResNet-50 metric learning · MSMT17 | mAP: 44.5% -> 46.3% | Improved |
| replace batch-wise OT loss with MVP matching loss | ResNet-50 metric learning · Duke-MTMC | mAP: 66.8% -> 70.0% | Improved |
| replace batch-wise OT loss with MVP matching loss | ResNet-50 metric learning · CUHK03-Labeled | Top-1: 92.5% -> 93.7% | Improved |
| replace identification loss with batch-hard triplet loss | ResNet-50 · Market-1501 | mAP: 70.4% -> 76.9% | Improved |
| replace identification loss with batch-all triplet loss | ResNet-50 · MSMT17 | mAP: 33.9% -> 27.1% | Hurt |
| replace batch-wise OT loss with MVP matching loss | ResNet-50 metric learning · CUHK03-Detected | Top-5: 98.8% -> 98.8% | No clear effect |
| replace BH triplet, lifted and OT losses with MVP matching loss | ResNet-50 metric learning · Market-1501, Duke-MTMC, MSMT17, CUHK03-Labeled | convergence speed (mAP / top-1 vs epochs) | Improved |
Closest papers by meaning
- Graph Sampling Based Deep Metric Learning for Generalizable Person Re-Identification (CVPR 2022 · similarity 0.88)
- Hard-Aware Point-to-Set Deep Metric for Person Re-identification (ECCV 2018 · similarity 0.88)
- A Systematic Evaluation and Benchmark for Person Re-Identification: Features, Metrics, and Datasets (TPAMI 2018 · similarity 0.88)
- Embedding Deep Metric for Person Re-identification: A Study Against Large Variations (ECCV 2016 · similarity 0.88)
- Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification (ICCV 2017 · similarity 0.88)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · similarity 0.86)
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.
