CrossEarth-Gate: Fisher-Guided Adaptive Tuning Engine for Efficient Adaptation of Cross-Domain Remote Sensing Semantic Segmentation

2025-11-25

AI Summary

Existing parameter-efficient fine-tuning (PEFT) methods fail in remote sensing cross-domain semantic segmentation due to coupled spatial, semantic, and frequency domain shifts. To address this, we propose a Fisher-guided adaptive tuning engine. Our key contributions are: (1) the first modular toolbox explicitly designed for multi-dimensional domain shifts in remote sensing, comprising dedicated adapters for spatial, semantic, and frequency domains; (2) a Fisher information-based gradient contribution assessment mechanism that enables dynamic selection of critical modules and fine-grained gradient modulation; and (3) integration of multi-scale feature modeling with hierarchical activation strategies. Evaluated on 16 cross-domain remote sensing segmentation benchmarks, our method achieves state-of-the-art performance, significantly improving adaptation efficiency and generalization capability while maintaining parameter efficiency.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

In Remote Sensing (RS), Parameter-Efficient Fine-Tuning (PEFT) has emerged as a key approach to activate the generalizable representation ability of foundation models for downstream tasks. However, existing specialized PEFT methods often fail when applied to large-scale Earth observation tasks, as they are unable to fully handle the multifaceted and unpredictable domain gaps (eg, spatial, semantic, and frequency shifts) inherent in RS data. To overcome this, we propose CrossEarth-Gate, which introduces two primary contributions. First, we establish a comprehensive RS module toolbox to address multifaceted domain gaps, comprising spatial, semantic, and frequency modules. Second, we develop a Fisher-guided adaptive selection mechanism that operates on this toolbox. This selection is guided by Fisher Information to quantify each module's importance by measuring its contribution to the task-specific gradient flow. It dynamically activates only the most critical modules at the appropriate layers, guiding the gradient flow to maximize adaptation effectiveness and efficiency. Comprehensive experiments validate the efficacy and generalizability of our method, where CrossEarth-Gate achieves state-of-the-art performance across 16 cross-domain benchmarks for RS semantic segmentation. The code of the work will be released.

Problem

• Addresses multifaceted domain gaps in remote sensing data
• Enables efficient adaptation of foundation models for Earth observation
• Dynamically selects optimal modules using Fisher-guided mechanism

Innovation

• Comprehensive toolbox addresses spatial semantic frequency gaps
• Fisher-guided mechanism quantifies module importance dynamically
• Adaptive selection activates critical modules for efficiency