Research map / Detection and segmentation / Open-set and few-shot detection
Few-Shot Segmentation via Cycle-Consistent Transformer
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 |
|---|---|---|---|
| replace the residual-block baseline with CyCTR | PASCAL-5i, ResNet-50, 1-shot · PASCAL-5i 1-shot | mIoU: 59.3 -> 63.5 | Helped |
| replace the residual-block baseline with CyCTR | PASCAL-5i, ResNet-50, 5-shot · PASCAL-5i 5-shot | mIoU: 62.0 -> 67.5 | Helped |
| replace the strongest listed baseline with CyCTR | COCO-20i, 1-shot · COCO-20i 1-shot | mIoU: 32.4 -> 40.3 | Helped |
| replace the strongest listed baseline with CyCTR | COCO-20i, 5-shot · COCO-20i 5-shot | mIoU: 38.5 -> 45.6 | Helped |
| add the self-alignment block (ablation) | CyCTR baseline on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 59.3 -> 62.5 | Helped |
| add cross-alignment with vanilla attention (ablation) | self-alignment model on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 62.5 -> 62.9 | Helped |
| replace vanilla attention with cycle-consistent attention (ablation) | self- and cross-alignment model on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 62.9 -> 62.6 | Hurt |
| use cycle-consistent attention instead of predicted aggregation (ablation) | CyCTR on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 62.6 -> 63.5 | Helped |
| add cycle-consistent attention for background support pixels (ablation) | foreground-only CyCTR on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 63.0 -> 63.5 | Helped |
| increase encoder count from 1 to 2 (ablation) | CyCTR on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 62.4 -> 63.5 | Helped |
| increase encoder count from 2 to 3 (ablation) | CyCTR on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 63.5 -> 63.7 | Helped |
| increase hidden dimension from 128 to 256 (ablation) | CyCTR on PASCAL-5i · PASCAL-5i 1-shot | mIoU: 63.5 -> 64.0 | 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)
- Bidirectional Reciprocative Information Communication for Few-Shot Semantic Segmentation (ICML 2024 · similarity 0.87)
- A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
- The Devil is in Low-Level Features for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.87)
- Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining (CVPR 2024 · similarity 0.86)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 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.