Research map / Detection and segmentation / Open-set and few-shot detection
Adaptive Agent Transformer for Few-Shot Segmentation
Yuan Wang, Rui Sun, Zhe Zhang, Tianzhu Zhang
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 the representation encoder (ablation) | Baseline · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 62.2 -> 64.9 | Improved |
| add the agent matching decoder (ablation) | Baseline + REnc · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 64.9 -> 67.4 | Improved |
| add the agent learning decoder (ablation) | Baseline + REnc + AMD · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 67.4 -> 69.1 | Improved |
| replace K-Means with agent learning decoder without OT (ablation) | DT-initialized agents · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 67.4 -> 68.3 | Improved |
| add optimal-transport equal partition constraint (ablation) | ALD · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 68.3 -> 69.1 | Improved |
| use distance-transform instead of learnable token initialization (ablation) | ALD with OT · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: 68.5 -> 69.1 | Improved |
| increase number of agent tokens K up to 14 (ablation) | AAFormer · Pascal-5i fold 0 1-shot, ResNet-50 | mIoU: performance grows until K = 14 | Improved |
| replace CyCTR with AAFormer (ResNet-50, 1-shot) | CyCTR · Pascal-5i 1-shot | mIoU: 64.2 -> 65.2 | Improved |
| replace CyCTR with AAFormer (ResNet-50, 5-shot) | CyCTR · Pascal-5i 5-shot | mIoU: 65.6 -> 67.6 | Improved |
| replace ASGNet with AAFormer (ResNet-101) | ASGNet · Pascal-5i 1-shot | mIoU: 59.3 -> 65.3 | Improved |
| replace CyCTR with AAFormer on COCO-20i (1-shot) | CyCTR · COCO-20i 1-shot | mIoU: 40.3 -> 41.6 | Improved |
| replace CyCTR with AAFormer on COCO-20i (5-shot) | CyCTR · COCO-20i 5-shot | mIoU: 45.6 -> 46.9 | Improved |
Closest papers by meaning
- Dual-Agent Optimization framework for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.89)
- Adaptive FSS: A Novel Few-Shot Segmentation Framework via Prototype Enhancement (AAAI 2024 · similarity 0.85)
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.85)
- Beyond Pixel and Object: Part Feature as Reference for Few-Shot Video Object Segmentation (AAAI 2025 · similarity 0.85)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.85)
- Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentation (CVPR 2024 · similarity 0.85)
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.
