Research map / Detection and segmentation / Open-set and few-shot detection
Meta-Tuning Loss Functions and Data Augmentation for Few-Shot Object Detection
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 |
|---|---|---|---|
| Meta-tune dynamic temperature, novel-class score scaling, and augmentation magnitude | MPSR · Pascal VOC novel classes, 1-shot | mAP: 33.1 -> 35.8 | Helped |
| Meta-tune dynamic temperature, novel-class score scaling, and augmentation magnitude | MPSR · Pascal VOC all classes, 1-shot | HM: 43.1 -> 46.3 | Helped |
| Meta-tune dynamic temperature, novel-class score scaling, and augmentation magnitude | MPSR · MS-COCO novel classes, 10-shot | mAP: 9.1 -> 12.5 | Helped |
| Meta-tune dynamic temperature, novel-class score scaling, and augmentation magnitude | DeFRCN · MS-COCO all classes, 30-shot | HM: 26.8 -> 28.0 | Helped |
| Add proxy-novel imitation (ablation) | MPSR+Meta-Dynamic · Pascal VOC Split-1, 5-shot, G-FSOD | HM: 61.5 -> 61.8 | Helped |
| Add model re-initialization for each task (ablation) | MPSR+Meta-Dynamic with proxy-novel imitation · Pascal VOC Split-1, 5-shot, G-FSOD | HM: 61.8 -> 62.1 | Helped |
| Add within-episode reward normalization (ablation) | MPSR+Meta-Dynamic with proxy-novel imitation and model re-initialization · Pascal VOC Split-1, 5-shot, G-FSOD | HM: 62.1 -> 63.3 | Helped |
| Add within-episode reward normalization (ablation) | MPSR+Meta-Dynamic · Pascal VOC Split-1, 5-shot, G-FSOD | 95% confidence interval: ± 0.75 -> ± 0.13 | Helped |
| Add photometric augmentation magnitude tuning only (ablation) | MPSR · Pascal VOC Split-1, 1-shot, novel classes | mAP: 33.1 -> 34.6 | Helped |
| Apply meta-tuned scaled dynamic temperature and augmentation | TFA · MS-COCO novel classes, 1-shot | mAP: 3.4 -> 4.7 | Helped |
Closest papers by meaning
- Frustratingly Simple Few-Shot Object Detection (ICML 2020 · similarity 0.94)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.91)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.90)
- A Closer Look at Cross-Domain Few-Shot Object Detection: Fine-Tuning Matters and Parallel Decoder Helps (CVPR 2026 · similarity 0.90)
- Strong Baselines for Parameter-Efficient Few-Shot Fine-Tuning (AAAI 2024 · similarity 0.88)
- Unleashing the Power of Meta-tuning for Few-shot Generalization Through Sparse Interpolated Experts (ICML 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.