Research map / Vision-language and multimodal / Human and embodied modeling
Dynamic Metric Learning: Towards a Scalable Metric Space To Accommodate Multiple Semantic Scales
Yifan Sun, Yuke Zhu, Yuhan Zhang, Pengkun Zheng, Xi Qiu, Chi Zhang, Yichen Wei
Check 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 |
|---|---|---|---|
| Apply Cross-Scale Learning instead of CosFace | DyML-Animal · DyML-Animal | mAP: 40.3 -> 45.8 | Improved |
| Apply Cross-Scale Learning instead of CosFace | DyML-Product · DyML-Product | mAP: 26.3 -> 37.0 | Improved |
| Apply Cross-Scale Learning instead of CosFace | DyML-Animal · DyML-Animal fine, middle, and coarse | mAP: 9.7 -> 12.7; 42.9 -> 50.3; 68.2 -> 74.4 | Improved |
| Apply Cross-Scale Learning instead of CosFace | DyML-Product · DyML-Product fine, middle, and coarse | mAP: 31.1 -> 38.0; 10.8 -> 17.9; 37.1 -> 55.1 | Improved |
| Use pairwise training rather than classification training with CSL (ablation) | CSL · DyML-Vehicle | mAP: 77.5 -> 53.3 | Hurt |
| Add joint classification and pairwise training to CSL (ablation) | CSL with classification training · DyML-Vehicle | mAP: 77.5 -> 79.1 | Improved |
| Add joint classification and pairwise training to CSL (ablation) | CSL with classification training · DyML-Vehicle | R@1: 92.6 -> 94.4 | Improved |
| Use CSL rather than the softmax multi-scale baseline | DyML-Vehicle training · DyML-Vehicle fine-scale training | classification accuracy: 80.3% -> 84.1% | Improved |
| Use multi-scale supervision rather than fine-only supervision | CosFace · DyML-Vehicle overall | Rank-1 accuracy: +3.8% | Improved |
| Use multi-scale supervision rather than middle-only supervision | CosFace · DyML-Vehicle overall | Rank-1 accuracy: +22.1% | Improved |
| Use multi-scale supervision rather than coarse-only supervision | CosFace · DyML-Vehicle overall | Rank-1 accuracy: +44.8% | Improved |
Closest papers by meaning
- Deep Metric Learning with Adaptively Composite Dynamic Constraints (TPAMI 2023 · similarity 0.88)
- MIC: Mining Interclass Characteristics for Improved Metric Learning (ICCV 2019 · similarity 0.87)
- DIML: Deep Interpretable Metric Learning via Structural Matching (TPAMI 2023 · similarity 0.87)
- Improving Deep Metric Learning by Divide and Conquer (TPAMI 2021 · similarity 0.87)
- Towards Interpretable Deep Metric Learning With Structural Matching (ICCV 2021 · similarity 0.87)
- Learning Semantic Proxies from Visual Prompts for Parameter-Efficient Fine-Tuning in Deep Metric Learning (ICLR 2024 · 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.
