Research map / Detection and segmentation / Open-set and few-shot detection
Attention-Based Multi-Context Guiding for Few-Shot Semantic Segmentation
Read the paperCheck 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 multi-context guiding (fuse Res-2..5 features of support and query) (ablation) | ResNet101 baseline · PASCAL-5i 1-shot | mIoU: 53.0 -> 55.3 | Helped |
| fuse context-2345 instead of only Res-5 (ablation) | MCG · PASCAL-5i sub-0 1-shot | mIoU: 61.2 -> 63.6 | Helped |
| add shared spatial attention (RAM) (ablation) | MCG · PASCAL-5i 1-shot | mIoU: 55.3 -> 57.3 | Helped |
| use shared spatial attention instead of separate (ablation) | MCG · PASCAL-5i sub-0 1-shot | mIoU: 63.3 -> 65.8 | Helped |
| use channel attention (SE) shared (ablation) | MCG · PASCAL-5i sub-0 1-shot | mIoU: 63.3 -> 61.7 | Hurt |
| add Conv-LSTM multi-shot fusion with per-shot loss (ablation) | A-MCG · PASCAL-5i 5-shot | mIoU: 57.8 -> 62.2 | Helped |
| supervise only the last shot (1-loss) instead of all shots (ablation) | Conv-LSTM · PASCAL-5i sub-0 1-shot | mIoU: 66.1 -> 65.1 | Hurt |
| change fusion width from 256 to 1024 (ablation) | MCG · PASCAL-5i sub-0 1-shot | mIoU: 63.6 -> 63.2 | Hurt |
| change fusion width from 256 to 64 (ablation) | MCG · PASCAL-5i sub-0 1-shot | mIoU: 63.6 -> 63.4 | No clear effect |
| full A-MCG-Conv-LSTM vs baseline | ResNet101 baseline · COCO 5-shot | mIoU: 51.2 -> 54.7 | Helped |
Closest papers by meaning
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.89)
- Unleashing the Potential of the Diffusion Model in Few-shot Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
- Visual Prompting for Generalized Few-shot Segmentation: A Multi-scale Approach (CVPR 2024 · similarity 0.86)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.86)
- APSeg: Auto-Prompt Network for Cross-Domain Few-Shot Semantic Segmentation (CVPR 2024 · similarity 0.86)
- Addressing Background Context Bias in Few-Shot Segmentation through Iterative Modulation (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.