Research map / Detection and segmentation / Open-set and few-shot detection
Few-Shot Object Detection With Attention-RPN and Multi-Relation Detector
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 |
|---|---|---|---|
| Train on FSOD rather than COCO | Few-shot detector · MS COCO minival, 20 novel categories, 10-shot | AP: 11.1 -> 16.6 | Helped |
| Fine-tune the FSOD-trained model on ImageNet Detection | Few-shot detector · ImageNet Detection, 50-way 5-shot | AP50: 41.7 -> 44.1 | Helped |
| Train with more categories while keeping a similar number of images (ablation) | Few-shot detector trained on FSOD · FSOD test, 5-way 5-shot | AP50: 60.3 -> 62.7 -> 64.7 | Helped |
| Add Attention-RPN (ablation) | Few-shot detector · FSOD test, 1-way 1-shot training | AP50: 55.0 -> 55.9 | Helped |
| Add Attention-RPN (ablation) | Few-shot detector · FSOD test, 2-way 5-shot training | AP50: 65.4 -> 67.5 | Helped |
| Use 2-way rather than 1-way contrastive training (ablation) | Few-shot detector · FSOD test | AP50: 55.0 -> 63.8 | Helped |
| Increase training support examples from 1-shot to 5-shot (ablation) | 2-way few-shot detector · FSOD test | AP50: 63.8 -> 65.4 | Helped |
| Use 5-way rather than 2-way training (ablation) | Few-shot detector with Attention-RPN and 5-shot supports · FSOD test | AP50: 67.5 -> 66.9 | Hurt |
| Use all three relation heads rather than any single head (ablation) | Multi-relation detector · FSOD test, 1-way 1-shot training | AP50: 47.7; 50.5; 45.1 -> 55.0 | Helped |
| Use the patch relation head alone (ablation) | Multi-relation detector · FSOD test, 1-way 1-shot training | AP50: 45.1 | Hurt |
Closest papers by meaning
- Restoring Negative Information in Few-Shot Object Detection (NeurIPS 2020 · similarity 0.90)
- Meta Faster R-CNN: Towards Accurate Few-Shot Object Detection with Attentive Feature Alignment (AAAI 2022 · similarity 0.89)
- Generalized Few-Shot Object Detection Without Forgetting (CVPR 2021 · similarity 0.89)
- Few-Shot Object Detection via Classification Refinement and Distractor Retreatment (CVPR 2021 · similarity 0.89)
- Few-Shot Object Detection with Foundation Models (CVPR 2024 · similarity 0.88)
- Revisiting Few-Shot Object Detection with Vision-Language Models (NeurIPS 2024 · similarity 0.87)
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.