Research map / Detection and segmentation / Open-set and few-shot detection
FarSeg++: Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
Zhuo Zheng, Yanfei Zhong, Junjue Wang, Ailong Ma, Liangpei 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 F-S relation module (ablation) | FPN baseline · iSAID val | mIoU (%): 59.31 -> 60.42 | Improved |
| add scale-aware projection to F-S relation (ablation) | FPN baseline + F-S relation · iSAID val | mIoU (%): 60.42 -> 60.49 | Improved |
| add F-A optimization (ablation) | FPN baseline · iSAID val | mIoU (%): 59.31 -> 61.51 | Improved |
| add F-A optimization (ablation) | baseline + F-S relation + scale-aware projection · iSAID val | mIoU (%): 60.49 -> 63.71 | Improved |
| add F-A optimization to F-S relation without scale-aware projection (ablation) | baseline + F-S relation · iSAID val | mIoU (%): 60.42 -> 63.21 | Improved |
| add scale-aware projection when F-A optimization is used (ablation) | baseline + F-S relation + F-A opt. · iSAID val | mIoU (%): 63.21 -> 63.71 | Improved |
| replace cross entropy with naive (1-p)^gamma weighted loss (ablation) | FarSeg · iSAID val | mIoU (%): 60.49 -> 56.44 | Hurt |
| add normalization constant Z to the weighted loss (ablation) | naive weighted loss · iSAID val | mIoU (%): 56.44 -> 62.98 | Improved |
| add linear annealing to dynamic weighting (ablation) | normalized F-A optimization · iSAID val | mIoU (%): 62.98 -> 63.18 | Improved |
| add poly annealing to dynamic weighting (ablation) | normalized F-A optimization · iSAID val | mIoU (%): 62.98 -> 63.52 | Improved |
| add cosine annealing to dynamic weighting (ablation) | normalized F-A optimization · iSAID val | mIoU (%): 62.98 -> 63.71 | Improved |
| increase focusing factor gamma from 0 to 2 (ablation) | FarSeg · iSAID val | mIoU (%): 60.42 -> 63.71 | Improved |
Closest papers by meaning
- Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery (CVPR 2020 · similarity 0.98)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.87)
- GeoViS: Geospatially Rewarded Visual Search for Remote Sensing Visual Grounding (CVPR 2026 · similarity 0.87)
- InstructSAM: A Training-free Framework for Instruction-Oriented Remote Sensing Object Recognition (NeurIPS 2025 · similarity 0.86)
- LISAt: Language-Instructed Segmentation Assistant for Satellite Imagery (NeurIPS 2025 · similarity 0.86)
- ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object (CVPR 2025 · 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.
