Research map / Detection and segmentation / Open-set and few-shot detection
Multi-Scale Positive Sample Refinement for Few-Shot Object Detection
Jiaxi Wu, Songtao Liu, Di Huang, Yunhong Wang
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 MPSR refinement branch | Baseline-FPN · VOC split 1, 1-shot | mAP (%): 25.5 -> 41.7 | Improved |
| add MPSR refinement branch | Baseline-FPN · VOC split 1, 10-shot | mAP (%): 56.9 -> 61.8 | Improved |
| add MPSR refinement branch | Baseline-FPN · COCO minival 30-shot | AP (%): 13.7 -> 14.1 | Improved |
| add MPSR refinement branch | Baseline-FPN · COCO minival 10-shot | AP (%): 9.5 -> 9.8 | Improved |
| add MPSR refinement branch | Baseline-FPN · COCO to VOC cross-dataset 10-shot | mAP (%): 39.3 -> 42.3 | Improved |
| add FPN to Faster R-CNN | Baseline · VOC split 1, 10-shot | mAP (%): 47.9 -> 56.9 | Improved |
| add FPN to Faster R-CNN | Baseline · VOC split 1, 1-shot | mAP (%): 24.5 -> 25.5 | No clear effect |
| use manual feature-level selection instead of anchor matching on object pyramids (ablation) | MPSR · VOC split 1, 1-shot | mAP (%): 30.8 -> 41.7 | Improved |
| add RoI refinement to RPN-only refinement (ablation) | MPSR · VOC split 1, 1-shot | mAP (%): 36.7 -> 41.7 | Improved |
| add RPN refinement to RoI-only refinement (ablation) | MPSR · VOC split 1, 1-shot | mAP (%): 33.7 -> 41.7 | Improved |
| apply refinement in both base training and fine-tuning instead of only few-shot (ablation) | MPSR · VOC split 1, 1-shot | mAP (%): 36.5 -> 41.7 | Improved |
| apply refinement in both steps instead of only base training (ablation) | MPSR · VOC split 1, 1-shot | mAP (%): 26.6 -> 41.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SNIPER — SNIPER: Efficient Multi-Scale Training (NeurIPS 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Few-Shot Object Detection via Classification Refinement and Distractor Retreatment (CVPR 2021 · similarity 0.92)
- Label, Verify, Correct: A Simple Few Shot Object Detection Method (CVPR 2022 · similarity 0.92)
- Few-Shot Object Detection with Foundation Models (CVPR 2024 · similarity 0.90)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.90)
- Multi-Perspective Data Augmentation for Few-shot Object Detection (ICLR 2025 · similarity 0.89)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 2024 · similarity 0.89)
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.
