Research map / Detection and segmentation / Open-set and few-shot detection
Accurate Few-Shot Object Detection With Support-Query Mutual Guidance and Hybrid Loss
Check 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 support-guided dynamic kernel query enhancement (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 15.6 -> 18.8 | Helped |
| Add class-agnostic proposal refinement (ablation) | Few-shot detector with kernel generator · COCO minival, 20 novel classes, 10-shot | AP75: 4.0 -> 7.4 | Helped |
| Add multi-level proposal scoring (ablation) | Few-shot detector with CAR · COCO minival, 20 novel classes, 10-shot | AP50: 19.3 -> 23.1 | Helped |
| Replace binary cross-entropy with the hybrid loss (ablation) | Few-shot detector with MPS · COCO minival, 20 novel classes, 10-shot | AP50: 23.1 -> 26.5 | Helped |
| Add query-guided support weighting (ablation) | Few-shot detector · COCO, 10-shot | AP50: 26.5 -> 27.0 | Helped |
| Use basic contrastive loss instead of the hybrid loss (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 27.0 -> 18.0 | Hurt |
| Add adaptive margin to basic contrastive loss (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 18.0 -> 18.7 | Helped |
| Use downsampling for negative proposals with adaptive margin (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 18.7 -> 24.4 | Helped |
| Use OHEM for negative proposals with adaptive margin (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 18.7 -> 25.1 | Helped |
| Replace adaptive margin plus OHEM with the hybrid loss (ablation) | Few-shot detector · COCO minival, 20 novel classes, 10-shot | AP50: 25.1 -> 27.0 | Helped |
| Score proposals at all FPN levels instead of only the top level (ablation) | Few-shot detector · COCO, 10-shot | AP: 10.3 -> 12.6 | Helped |
| Score proposals at all FPN levels instead of a single size-selected level (ablation) | Few-shot detector · COCO, 10-shot | AP: 11.5 -> 12.6 | Helped |
Closest papers by meaning
- Kernelized Few-Shot Object Detection With Efficient Integral Aggregation (CVPR 2022 · similarity 0.89)
- Label, Verify, Correct: A Simple Few Shot Object Detection Method (CVPR 2022 · similarity 0.89)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.87)
- Few-Shot Object Detection with Foundation Models (CVPR 2024 · similarity 0.87)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 2024 · similarity 0.86)
- SNIDA: Unlocking Few-Shot Object Detection with Non-linear Semantic Decoupling Augmentation (CVPR 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.