Research map / Detection and segmentation / Open-set and few-shot detection
AgMTR: Agent Mining Transformer for Few-Shot Segmentation in Remote Sensing
Hanbo Bi, Yingchao Feng, Yongqiang Mao, Jianning Pei, Wenhui Diao, 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 Agent Learning Encoder (ALE) to baseline (ablation) | AgMTR baseline (DeiT-B/16) · iSAID 1-shot | mIoU: 46.82 -> 48.94 | Improved |
| add Agent Aggregation Decoder (AAD) on top of ALE (ablation) | AgMTR · iSAID 1-shot | mIoU: 48.94 -> 50.01 | Improved |
| add Semantic Alignment Decoder (SAD) on top of ALE (ablation) | AgMTR · iSAID 1-shot | mIoU: 48.94 -> 50.27 | Improved |
| add SAD on top of ALE+AAD (ablation) | AgMTR · iSAID 1-shot | mIoU: 50.01 -> 51.58 | Improved |
| add all three components to baseline (ablation) | AgMTR · iSAID 1-shot | mIoU: 46.82 -> 51.58 | Improved |
| add support-prototype initialization of agents (ablation) | AgMTR ALE · iSAID 1-shot | mIoU: 49.12 -> 51.58 | Improved |
| add optimal transport local masks (ablation) | AgMTR ALE · iSAID 1-shot | mIoU: 50.74 -> 51.58 | Improved |
| add GAT in SAD (ablation) | AgMTR SAD · iSAID 1-shot | mIoU: 50.89 -> 51.58 | Improved |
| increase unlabeled images from 0 to 5 (ablation) | AgMTR AAD · iSAID 1-shot | mIoU: 50.27 -> 51.58 | Improved |
| increase unlabeled images from 5 to 9 (ablation) | AgMTR AAD · iSAID 1-shot | mIoU: 51.58 -> 51.61 | No clear effect |
| use 1 unlabeled image instead of 5 (ablation) | AgMTR AAD · iSAID 1-shot | mIoU: 50.64 -> 51.58 | Improved |
| add all agents' semantic sources (support, unlabeled, query) (ablation) | AgMTR · iSAID (cosine distance) | cosine distance to query prototype: 0.5843 -> 0.7465 | Improved |
Closest papers by meaning
- Dual-Agent Optimization framework for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.85)
- Beyond Confidence: Exploiting Homogeneous Pattern for Semi-Supervised Semantic Segmentation (ICML 2025 · similarity 0.85)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.84)
- LISAt: Language-Instructed Segmentation Assistant for Satellite Imagery (NeurIPS 2025 · similarity 0.84)
- Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agents (ICLR 2026 · similarity 0.84)
- ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object (CVPR 2025 · similarity 0.83)
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.
