Research map / Label-efficient and robust learning / Recognition and long-tail
Mancs: A Multi-task Attentional Network with Curriculum Sampling for Person Re-identification
Cheng Wang, Qian Zhang, Chang Huang, Wenyu Liu, Xinggang Wang
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 random erasing (ablation) | global classification baseline · Market-1501 (single query) | mAP and rank1: rank1 69.5 -> 71.6; mAP 46.1 -> 47.6 | Improved |
| add random erasing (ablation) | global classification baseline · CUHK03 detected (767/700 split) | mAP and rank1: rank1 33.9 -> 42.2; mAP 30.8 -> 38.9 | Improved |
| add triplet ranking branch with OHEM (ablation) | classification + random erasing · Market-1501 (single query) | mAP and rank1: rank1 71.6 -> 92.4; mAP 47.6 -> 80.4 | Improved |
| add triplet ranking branch with OHEM (ablation) | classification + random erasing · CUHK03 detected (767/700 split) | mAP and rank1: rank1 42.2 -> 63.8; mAP 38.9 -> 58.4 | Improved |
| use focal loss instead of cross-entropy (ablation) | Mancs with OHEM · Market-1501 (single query) | mAP and rank1: rank1 92.4 -> 92.7; mAP 80.4 -> 81.0 | Improved |
| use focal loss instead of cross-entropy (ablation) | Mancs with OHEM · CUHK03 detected (767/700 split) | mAP and rank1: rank1 63.8 -> 63.9; mAP 58.4 -> 59.2 | Improved |
| add fully attentional block with deep supervision (ablation) | Mancs with OHEM and focal loss · Market-1501 (single query) | mAP and rank1: rank1 92.7 -> 92.9; mAP 81.0 -> 81.7 | Improved |
| add fully attentional block with deep supervision (ablation) | Mancs with OHEM and focal loss · CUHK03 detected (767/700 split) | mAP and rank1: rank1 63.9 -> 64.4; mAP 59.2 -> 60.1 | Improved |
| use curriculum sampling instead of OHEM (ablation) | Mancs full · Market-1501 (single query) | mAP and rank1: rank1 92.9 -> 93.1; mAP 81.7 -> 82.3 | Improved |
| use curriculum sampling instead of OHEM (ablation) | Mancs full · CUHK03 detected (767/700 split) | mAP and rank1: rank1 64.4 -> 65.5; mAP 60.1 -> 60.5 | Improved |
| add re-ranking post-processing | Mancs · Market-1501 (single query) | mAP and rank1: mAP 82.3 -> 92.3; rank1 93.1 -> 94.9 | Improved |
| add re-ranking post-processing | Mancs · Market-1501 (multiple query) | mAP and rank1: mAP 87.5 -> 94.5; rank1 95.4 -> 95.8 | Improved |
Later papers that compared with this method
4 later accepted papers report Mancs as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Salience-Guided Cascaded Suppression Network for Person Re-Identification (CVPR 2020 · better on 0 of 2 shared results)
- Interaction-And-Aggregation Network for Person Re-Identification (CVPR 2019 · better on 3 of 3 shared results)
- Joint Discriminative and Generative Learning for Person Re-Identification (CVPR 2019 · better on 3 of 3 shared results)
- Densely Semantically Aligned Person Re-Identification (CVPR 2019 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Harmonious Attention Network for Person Re-Identification (CVPR 2018 · similarity 0.91)
- Scalable Person Re-Identification by Harmonious Attention (IJCV 2019 · similarity 0.91)
- Leader-Based Multi-Scale Attention Deep Architecture for Person Re-Identification (TPAMI 2019 · similarity 0.90)
- Deeply-Learned Part-Aligned Representations for Person Re-Identification (ICCV 2017 · similarity 0.89)
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.89)
- Resource Aware Person Re-Identification Across Multiple Resolutions (CVPR 2018 · similarity 0.89)
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.
