Research map / Label-efficient and robust learning / Domain adaptation and generalization
CrossEarth-Gate: Fisher-Guided Adaptive Tuning Engine for Efficient Adaptation of Cross-Domain Remote Sensing Semantic Segmentation
Shilei Cao, Ziyang Gong, Hehai Lin, Yang Liu, Jiashun Cheng, Xiaoxing Hu, Haoyuan Liang, Guowen Li, Chengwei Qin, Hong Cheng, Xue Yang, Juepeng Zheng and 1 more
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 Fisher-guided selection of modules (vs training all toolbox modules) (ablation) | CrossEarth-Gate · CASID DG (mean over 4 source domains) | mIoU: 59.6 -> 61.9 | Improved |
| add spatial (LoRA) modules to toolbox (ablation) | CrossEarth-Gate · CASID DG (mean over 4 source domains) | mIoU: 60.5 -> 61.9 | Improved |
| add semantic (adapter) modules to toolbox (ablation) | CrossEarth-Gate · CASID DG (mean over 4 source domains) | mIoU: 60.4 -> 61.9 | Improved |
| add frequency (Earth-Adapter) modules to toolbox (ablation) | CrossEarth-Gate · CASID DG (mean over 4 source domains) | mIoU: 60.4 -> 61.9 | Improved |
| full fine-tuning instead of frozen backbone | DINOv2-L · CASID DG (source Tem, avg) | mIoU: 63.6 -> 52.8 | Hurt |
| full fine-tuning instead of frozen backbone | DINOv2-Base · CASID DG (mean over 4 source domains) | mIoU: 55.3 -> 50.9 | Hurt |
| apply CrossEarth-Gate to different backbones | DINOv2-Base · CASID DG (mean over 4 source domains) | mIoU: 55.3 -> 57.8 | Improved |
| apply CrossEarth-Gate to SatMAE-Large | SatMAE-Large · CASID DG (mean over 4 source domains) | mIoU: 24.3 -> 31.9 | Improved |
| CrossEarth-Gate vs Earth-Adapter on Tem2Sub | CrossEarth-Gate · CASID DG (Tem2Sub) | mIoU: 68.8 -> 68.0 | Hurt |
| CrossEarth-Gate vs Frozen on Building class | CrossEarth-Gate · P(r)2Res | IoU (Building): 61.7 -> 58.6 | Hurt |
| CrossEarth-Gate vs static PEFT in DA | CrossEarth-Gate · DA average (P2V, V2P, R2U, U2R) | mIoU: 58.1 -> 59.1 | Improved |
| full fine-tuning instead of frozen in DA with pseudo-labels | DINOv2-L · DA average (P2V, V2P, R2U, U2R) | mIoU: 56.4 -> 50.1 | Hurt |
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.
- Earth-Adapter — Earth-Adapter: Bridge the Geospatial Domain Gaps with a Frequency-Guided Mixture of Adapters (AAAI 2026) · better on 6 of 6 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
- CrossEarth: Geospatial Vision Foundation Model for Domain Generalizable Remote Sensing Semantic Segmentation (TPAMI 2025 · similarity 0.92)
- Earth-Adapter: Bridge the Geospatial Domain Gaps with a Frequency-Guided Mixture of Adapters (AAAI 2026 · similarity 0.90)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.87)
- Meta-Learning Hyperparameters for Parameter Efficient Fine-Tuning (CVPR 2025 · similarity 0.87)
- ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object (CVPR 2025 · similarity 0.87)
- SemStereo: Semantic-Constrained Stereo Matching Network for Remote Sensing (AAAI 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.
