Research map / Federated learning and privacy / Personalized and heterogeneous FL
Capturing Spectral and Spatial Patterns for Federated Remote Sensing Segmentation
Yixin Xue, Wenke Huang, Haonan Guo, Bo Du
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 invariant-prototype supervision (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(RGB) (spectral+spatial shift) | mIoU: 68.35 -> 70.42 | Improved |
| add invariant-prototype supervision (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(IRRG) (spatial shift) | mIoU: 72.10 -> 73.43 | Improved |
| add invariant-prototype supervision (ablation) | FedDap · LoveDA urban/rural (spatial shift) | mIoU: 50.38 -> 50.68 | Improved |
| add DAP-based (dynamic prototype) supervision (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(RGB) (spectral+spatial shift) | mIoU: 69.05 -> 70.42 | Improved |
| add DAP-based (dynamic prototype) supervision (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(IRRG) (spatial shift) | mIoU: 72.44 -> 73.43 | Improved |
| add spatial branch to spectral-only (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(IRRG) (spatial shift) | mIoU: 72.18 -> 73.43 | Improved |
| add spectral branch to spatial-only (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(IRRG) (spatial shift) | mIoU: 73.31 -> 73.43 | No clear effect |
| add both branches rather than a single one (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(RGB) (spectral+spatial shift) | mIoU: 69.41 -> 70.42 | Improved |
| use learned shift-aware gate instead of fixed g=0.5 (ablation) | FedDap · LoveDA urban/rural (spatial shift) | mIoU: 49.83 -> 50.68 | Improved |
| use learned gate instead of cascaded serial fusion (ablation) | FedDap · Potsdam(IRRG) vs Vaihingen(IRRG) (spatial shift) | mIoU: 69.75 -> 73.43 | Improved |
| use FedDap instead of FedSeg | FedSeg · LoveDA urban/rural (spatial shift) | mIoU: 50.04 -> 50.68 | Improved |
| use FedDap instead of FedSeg | FedSeg · Potsdam(IRRG) vs Vaihingen(RGB) (spectral+spatial shift) | mIoU: 68.82 -> 70.42 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FedSeg — FedSeg: Class-Heterogeneous Federated Learning for Semantic Segmentation (CVPR 2023) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- From Visual Primitives to Semantic Masks: Fine-Grained Visual-Linguistic Alignment for Open-Vocabulary Remote Sensing Image Segmentation (ECCV 2026 · similarity 0.87)
- FedSeg: Class-Heterogeneous Federated Learning for Semantic Segmentation (CVPR 2023 · similarity 0.87)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.86)
- SemStereo: Semantic-Constrained Stereo Matching Network for Remote Sensing (AAAI 2025 · similarity 0.85)
- Federated Black-Box Adaptation for Semantic Segmentation (NeurIPS 2024 · similarity 0.84)
- ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object (CVPR 2025 · similarity 0.84)
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.
