Research map / Detection and segmentation / Open-set and few-shot detection
Prompt-and-Transfer: Dynamic Class-Aware Enhancement for Few-Shot Segmentation
Hanbo Bi, Yingchao Feng, Wenhui Diao, Peijin Wang, Yongqiang Mao, Kun Fu, Hongqi Wang, Xian Sun
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 FG-prompts (ablation) | PAT baseline · PASCAL-5i 1-shot | mIoU: 67.30 -> 68.26 | Improved |
| add FG-prompts (ablation) | PAT baseline · COCO-20i 1-shot | mIoU: 42.14 -> 43.75 | Improved |
| add Semantic Prompt Transfer (ablation) | PAT with FG-prompts · PASCAL-5i 1-shot | mIoU: 68.26 -> 69.57 | Improved |
| add BG-prompts (ablation) | PAT with FG-prompts and SPT · PASCAL-5i 1-shot | mIoU: 69.57 -> 70.52 | Improved |
| add Part Mask Generator (ablation) | PAT with FG, BG and SPT · PASCAL-5i 1-shot | mIoU: 70.52 -> 71.66 | Improved |
| add Part Mask Generator (ablation) | PAT with FG, BG and SPT · COCO-20i 1-shot | mIoU: 46.71 -> 48.05 | Improved |
| initialize FG-prompts with CLIP text rather than random (ablation) | PAT FG-prompts · PASCAL-5i 1-shot | mIoU: 67.34 -> 68.26 | Improved |
| transfer semantics from both support and query instead of support only (ablation) | PAT SPT · PASCAL-5i 1-shot | mIoU: 71.07 -> 71.66 | Improved |
| add Gaussian suppression in SPT (ablation) | PAT SPT · PASCAL-5i 1-shot | mIoU: 71.37 -> 71.66 | Improved |
| increase number of prompt enhancements from 3 to 7 (ablation) | PAT · PASCAL-5i 1-shot | mIoU: 71.66 -> 71.79 | No clear effect |
| increase number of part masks beyond 8 (ablation) | PAT PMG · PASCAL-5i 1-shot | mIoU: below 71% | Hurt |
| use DeiT-B/16 instead of ViT-B/16 backbone | PAT · PASCAL-5i 1-shot | mIoU: 66.96 -> 71.66 | Improved |
Closest papers by meaning
- Visual Prompting for Generalized Few-shot Segmentation: A Multi-scale Approach (CVPR 2024 · similarity 0.89)
- PromptKD: Unsupervised Prompt Distillation for Vision-Language Models (CVPR 2024 · similarity 0.89)
- DSV-LFS: Unifying LLM-Driven Semantic Cues with Visual Features for Robust Few-Shot Segmentation (CVPR 2025 · similarity 0.88)
- SAM-Aware Graph Prompt Reasoning Network for Cross-Domain Few-Shot Segmentation (AAAI 2025 · similarity 0.88)
- APSeg: Auto-Prompt Network for Cross-Domain Few-Shot Semantic Segmentation (CVPR 2024 · similarity 0.88)
- Just a Few Glances: Open-Set Visual Perception with Image Prompt Paradigm (AAAI 2025 · similarity 0.88)
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.
