Research map / Label-efficient and robust learning / Recognition and long-tail
Group Consistent Similarity Learning via Deep CRF 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 |
|---|---|---|---|
| Add group-similarity supervision instead of local-similarity supervision (ablation) | Person re-identification model · Market-1501 | mAP: 70.6 -> 77.5 | Helped |
| Add group-similarity supervision instead of local-similarity supervision (ablation) | Person re-identification model · DukeMTMC-reID | mAP: 59.6 -> 64.8 | Helped |
| Use group-organized training batches rather than randomly shuffled pairs (ablation) | Three-scale local-similarity model · Market-1501 | mAP: 73.8 -> 78.4 | Helped |
| Use group-organized training batches rather than randomly shuffled pairs (ablation) | Three-scale local-similarity model · DukeMTMC-reID | mAP: 62.9 -> 67.8 | Helped |
| Add group supervision to equally averaged three-scale local similarities (ablation) | Three-scale person re-identification model · Market-1501 | mAP: 73.8 -> 80.5 | Helped |
| Add group supervision to equally averaged three-scale local similarities (ablation) | Three-scale person re-identification model · DukeMTMC-reID | mAP: 62.9 -> 68.0 | Helped |
| Add identification loss to the three-scale group model (ablation) | Three-scale group-similarity model · Market-1501 | mAP: 73.7 -> 81.6 | Helped |
| Add identification loss to the three-scale group model (ablation) | Three-scale group-similarity model · DukeMTMC-reID | mAP: 63.7 -> 69.5 | Helped |
| Replace equal averaging with CRF-learned weights for combining scales (ablation) | Three-scale group-similarity model · Market-1501 | mAP: 80.5 -> 81.6 | Helped |
| Replace equal averaging with CRF-learned weights for combining scales (ablation) | Three-scale group-similarity model · DukeMTMC-reID | mAP: 68.0 -> 69.5 | Helped |
| Use three-scale embeddings and equally average scale similarities instead of scale-1 group similarity (ablation) | Group-similarity model · Market-1501 | mAP: 78.7 -> 80.5 | Helped |
| Use three-scale embeddings and equally average scale similarities instead of scale-1 group similarity (ablation) | Group-similarity model · DukeMTMC-reID | mAP: 66.4 -> 68.0 | Helped |
Closest papers by meaning
- Efficient and Deep Person Re-Identification Using Multi-Level Similarity (CVPR 2018 · similarity 0.90)
- Consistent-Aware Deep Learning for Person Re-Identification in a Camera Network (CVPR 2017 · similarity 0.90)
- Point to Set Similarity Based Deep Feature Learning for Person Re-Identification (CVPR 2017 · similarity 0.89)
- Deep Neural Networks with Inexact Matching for Person Re-Identification (NeurIPS 2016 · similarity 0.89)
- Intra-Inter Camera Similarity for Unsupervised Person Re-Identification (CVPR 2021 · 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.