Research map / Detection and segmentation / Open-set and few-shot detection
Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
Check 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 |
|---|---|---|---|
| Use FarSeg instead of PSPNet | iSAID val segmentation · iSAID val | mIoU (%): 60.25 -> 63.71 | Helped |
| Add F-S relation to the FPN and light-decoder baseline (ablation) | FarSeg baseline · iSAID val | mIoU (%): 59.31 -> 60.42 | Helped |
| Add scale-aware projection to F-S relation (ablation) | FarSeg baseline with F-S relation · iSAID val | mIoU (%): 60.42 -> 60.49 | Helped |
| Add F-A optimization to the baseline (ablation) | FarSeg baseline · iSAID val | mIoU (%): 59.31 -> 61.51 | Helped |
| Add F-A optimization to the full F-S relation and scale-aware projection variant (ablation) | FarSeg · iSAID val | mIoU (%): 60.49 -> 63.71 | Helped |
| Normalize the (1-p)^γ loss weights (ablation) | F-A optimization · iSAID val | mIoU (%): 56.44 -> 62.98 | Helped |
| Use (1-p)^γ weighting without normalization (ablation) | F-A optimization · iSAID val | mIoU (%): 60.49 -> 56.44 | Hurt |
| Add linear annealing to normalized loss weighting (ablation) | F-A optimization · iSAID val | mIoU (%): 62.98 -> 63.18 | Helped |
| Use polynomial instead of linear annealing (ablation) | F-A optimization · iSAID val | mIoU (%): 63.18 -> 63.52 | Helped |
| Use cosine instead of polynomial annealing (ablation) | F-A optimization · iSAID val | mIoU (%): 63.52 -> 63.71 | Helped |
| Increase focusing factor from γ=0 to γ=2 (ablation) | F-A optimization · iSAID val | mIoU (%): 60.42 -> 63.71 | Helped |
| Increase focusing factor from γ=2 to γ=5 (ablation) | F-A optimization · iSAID val | mIoU (%): 63.71 -> 62.61 | Hurt |
Closest papers by meaning
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.88)
- GeoViS: Geospatially Rewarded Visual Search for Remote Sensing Visual Grounding (CVPR 2026 · similarity 0.87)
- EarthVQA: Towards Queryable Earth via Relational Reasoning-Based Remote Sensing Visual Question Answering (AAAI 2024 · similarity 0.87)
- SemStereo: Semantic-Constrained Stereo Matching Network for Remote Sensing (AAAI 2025 · similarity 0.87)
- LISAt: Language-Instructed Segmentation Assistant for Satellite Imagery (NeurIPS 2025 · similarity 0.87)
- InstructSAM: A Training-free Framework for Instruction-Oriented Remote Sensing Object Recognition (NeurIPS 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.