Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Meta Metric Learning
Guangyi Chen, Tianren Zhang, Jiwen Lu, Jie Zhou
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 hard sample mining on set-based distance (ablation) | DMML · Market-1501 | Rank-1: 87.1 -> 92.4 | Improved |
| add hard sample mining on set-based distance (ablation) | DMML · Market-1501 | mAP: 70.3 -> 81.0 | Improved |
| add additive margin on negative samples (ablation) | DMML · Market-1501 | Rank-1: 91.7 -> 92.4 | Improved |
| use negative-sample margin instead of positive-sample additive margin (AM) (ablation) | DMML · Market-1501 | Rank-1: 91.9 -> 92.4 | Improved |
| set margin tau to 0.4 instead of 0.2 (ablation) | DMML · Market-1501 | Rank-1: 92.0 -> 92.4 | No clear effect |
| set margin tau to 0.4 instead of 0.6 (ablation) | DMML · Market-1501 | Rank-1: 91.7 -> 92.4 | No clear effect |
| sample 32 classes per sub-task instead of 16 (ablation) | DMML · Market-1501 | Rank-1: 91.0 -> 92.4 | Improved |
| sample 64 classes per sub-task instead of 32 (ablation) | DMML · Market-1501 | Rank-1: 92.4 -> 92.2 | No clear effect |
| use 5 support samples per meta-cell instead of 1 (ablation) | DMML · Market-1501 | Rank-1: 90.0 -> 92.4 | Improved |
| use 7 support samples instead of 5 (ablation) | DMML · Market-1501 | mAP: 81.0 -> 81.6 | Improved |
| add part-based branch to DMML | DMML · Market-1501 | Rank-1: 92.4 -> 93.5 | Improved |
| add part-based branch to DMML | DMML · DukeMTMC-reID | mAP: 70.2 -> 73.7 | Improved |
Closest papers by meaning
- Deep Factorized Metric Learning (CVPR 2023 · similarity 0.87)
- Improving Deep Metric Learning by Divide and Conquer (TPAMI 2021 · similarity 0.87)
- Sharable and Individual Multi-View Metric Learning (TPAMI 2017 · similarity 0.86)
- DiVA: Diverse Visual Feature Aggregation for Deep Metric Learning (ECCV 2020 · similarity 0.86)
- Towards Interpretable Deep Metric Learning With Structural Matching (ICCV 2021 · similarity 0.86)
- Deep Metric Learning Beyond Binary Supervision (CVPR 2019 · 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.
