Research map / Detection and segmentation / Open-set and few-shot detection
FGN: Fully Guided Network for Few-Shot Instance Segmentation
Zhibo Fan, Jin-Gang Yu, Zhihao Liang, Jiarong Ou, Changxin Gao, Gui-Song Xia, Yuanqing Li
Check 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 |
|---|---|---|---|
| Use separate support-guidance modules across RPN, detector, and mask head (ablation) | FGN · COCO2VOC 3way-3shot | segmentation mAP50: 15.1 -> 17.9 | Improved |
| Use separate support-guidance modules across RPN, detector, and mask head (ablation) | FGN · COCO2VOC 3way-3shot | detection mAP50: 29.1 -> 32.9 | Improved |
| Add class-aware attention guidance to proposal generation (ablation) | FGN · COCO2VOC 3way-3shot | proposal AR50: 64.5 -> 92.5 | Improved |
| Use the proposed attention-guided mask head instead of unguided FCN (ablation) | FGN · COCO2VOC 3way-3shot | segmentation mAP50: 15.1 -> 17.9 | Improved |
| Replace the proposed AG-FCN with semantic-segmentation FCN guidance (ablation) | FGN · COCO2VOC 3way-3shot | segmentation mAP50: 17.9 -> 14.5 | Hurt |
| Tile attention vectors and concatenate them with RoI features (ablation) | AG-FCN · COCO2VOC 3way-3shot | segmentation mAP50: 17.9 -> 15.6 | Hurt |
| Use FGN rather than the strongest reported baseline | COCO2VOC · COCO2VOC 3way-3shot | segmentation mAP50: 15.3 -> 17.9 | Improved |
| Use FGN rather than the strongest reported baseline | COCO2VOC · COCO2VOC 3way-3shot | detection mAP50: 23.4 -> 32.9 | Improved |
| Use FGN rather than the strongest reported baseline | VOC2VOC · VOC2VOC 1way-1shot | detection mAP50: 17.9 -> 15.4 | Hurt |
| Use FGN rather than the strongest reported baseline | VOC2VOC · VOC2VOC 3way-1shot | segmentation mAP50: 3.6 -> 7.3 | Improved |
| Use FGN rather than the strongest reported baseline | VOC2VOC · VOC2VOC 3way-3shot | detection mAP50: 14.4 -> 16.2 | Improved |
| Use episodic support-conditioned FGN instead of fine-tuning Mask R-CNN | FSIS · COCO2VOC 3way-3shot | segmentation mAP50: 2.7 -> 17.9 | Improved |
Closest papers by meaning
- Reference Twice: A Simple and Unified Baseline for Few-Shot Instance Segmentation (TPAMI 2024 · similarity 0.89)
- Fully Convolutional Instance-Aware Semantic Segmentation (CVPR 2017 · similarity 0.88)
- Mask Encoding for Single Shot Instance Segmentation (CVPR 2020 · similarity 0.88)
- iFS-RCNN: An Incremental Few-Shot Instance Segmenter (CVPR 2022 · similarity 0.88)
- Instance-Sensitive Fully Convolutional Networks (ECCV 2016 · similarity 0.88)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (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.
