Research map / Detection and segmentation / Open-set and few-shot detection
Dense Cross-Query-and-Support Attention Weighted Mask Aggregation for Few-Shot Segmentation
Xinyu Shi, Dong Wei, Yu Zhang, Donghuan Lu, Munan Ning, Jiashun Chen, Kai Ma, Yefeng Zheng
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 |
|---|---|---|---|
| use background support information (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 67.3 -> 69.3 | Improved |
| use background support information (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | FB-IoU: 76.2 -> 78.5 | Improved |
| one-pass n-shot inference instead of voting over five 1-shot predictions (ablation) | DCAMA · PASCAL-5i 5-shot (Swin-B) | mIoU: 74.0 -> 74.9 | Improved |
| one-pass n-shot inference instead of SCL cross-guided averaging (ablation) | DCAMA · PASCAL-5i 5-shot (Swin-B) | mIoU: 74.1 -> 74.9 | Improved |
| one-pass n-shot inference instead of HSNet normalized voting (ablation) | DCAMA · PASCAL-5i 5-shot (Swin-B) | FB-IoU: 81.8 -> 82.9 | Improved |
| add skip connection of 1/4 and 1/8 scale features (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 66.2 -> 69.3 | Improved |
| add skip connection of 1/16 scale features (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 66.2 -> 59.8 | Hurt |
| add 1/4 and 1/8 and 1/16 skip connections instead of 1/4 and 1/8 only (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 69.3 -> 58.9 | Hurt |
| add 1/16 scale attention on top of 1/32 only (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 58.7 -> 67.2 | Improved |
| add 1/8 scale attention on top of 1/16 and 1/32 (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 67.2 -> 69.3 | Improved |
| skip connect support features in addition to query features (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 68.9 -> 69.3 | Improved |
| use all support pixels instead of 600 foreground and background pixels (ablation) | DCAMA · PASCAL-5i 1-shot (Swin-B) | mIoU: 65.7 -> 69.3 | Improved |
Closest papers by meaning
- Foreground-Covering Prototype Generation and Matching for SAM-Aided Few-Shot Segmentation (AAAI 2025 · similarity 0.90)
- Lightweight Frequency Masker for Cross-Domain Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.89)
- CGMGM: A Cross-Gaussian Mixture Generative Model for Few-Shot Semantic Segmentation (AAAI 2024 · similarity 0.88)
- DSV-LFS: Unifying LLM-Driven Semantic Cues with Visual Features for Robust Few-Shot Segmentation (CVPR 2025 · similarity 0.88)
- Adaptive FSS: A Novel Few-Shot Segmentation Framework via Prototype Enhancement (AAAI 2024 · similarity 0.88)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (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.
