Research map / Detection and segmentation / Open-set and few-shot detection
Universal-Prototype Enhancing for Few-Shot Object Detection
Aming Wu, Yahong Han, Linchao Zhu, Yi Yang
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 conditional (affine-transformed) prototypes instead of raw universal prototypes (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 43.8 -> 47.8 | Improved |
| add conditional (affine-transformed) prototypes instead of raw universal prototypes (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 1-shot | novel mAP (%): 38.1 -> 43.8 | Improved |
| use conditional prototypes instead of a new set of prototypes (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 44.6 -> 47.8 | Improved |
| use conditional prototypes instead of a new set of prototypes (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 5-shot | novel mAP (%): 56.1 -> 55.4 | Hurt |
| use 24 universal prototypes instead of 16 (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 42.7 -> 47.8 | Improved |
| use 24 universal prototypes instead of 32 (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 42.1 -> 47.8 | Improved |
| set consistency loss weight gamma to 1.0 instead of 0.8 (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 42.5 -> 47.8 | Improved |
| set consistency loss weight gamma to 1.0 instead of 1.4 (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 2-shot | novel mAP (%): 43.5 -> 47.8 | Improved |
| fuse prototype descriptors into RPN input (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 1-shot and 5-shot | novel mAP (%): +2.7% (1-shot), +1.8% (5-shot) | Improved |
| fuse prototype descriptors into classifier input (ablation) | FSOD-up · PASCAL VOC Novel Set 1, 1-shot and 5-shot | novel mAP (%): +2.1% (1-shot), +1.2% (5-shot) | Improved |
| use FSOD-up instead of MPSR | FSOD-up · PASCAL VOC Novel Set 2, 1-shot | novel mAP (%): 24.4 -> 31.2 | Improved |
| use FSOD-up instead of MPSR | FSOD-up · PASCAL VOC Novel Set 1, 5-shot vs TFA w/cos | novel mAP (%): 55.7 -> 55.4 | No clear effect |
Closest papers by meaning
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.91)
- Transformation Invariant Few-Shot Object Detection (CVPR 2021 · similarity 0.90)
- Learning General and Specific Embedding with Transformer for Few-Shot Object Detection (IJCV 2024 · similarity 0.89)
- Label, Verify, Correct: A Simple Few Shot Object Detection Method (CVPR 2022 · similarity 0.89)
- Few-Shot Object Detection with Foundation Models (CVPR 2024 · similarity 0.88)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 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.
