WILD-SAM: Phase-Aware Expert Adaptation of SAM for Landslide Detection in Wrapped InSAR Interferograms

arXiv cs.CV / 4/17/2026

📰 NewsDeveloper Stack & InfrastructureIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes WILD-SAM, a parameter-efficient fine-tuning method to adapt the Segment Anything Model (SAM) for detecting slow-moving landslides from wrapped InSAR interferograms despite severe phase ambiguity and coherence noise.
It addresses the spectral domain shift that harms boundary delineation by adding a Phase-Aware Mixture-of-Experts (PA-MoE) Adapter to align spectral distributions using dynamic routing over multi-scale convolutional experts.
To improve boundary quality and preserve topology, WILD-SAM introduces Wavelet-Guided Subband Enhancement (WGSE), which uses discrete wavelet transforms to isolate high-frequency subbands and convert phase textures into frequency-aware dense prompts.
Experiments on the ISSLIDE and ISSLIDE+ benchmarks show that WILD-SAM delivers state-of-the-art results, outperforming prior approaches on both target completeness and contour fidelity.

Abstract

Detecting slow-moving landslides directly from wrapped Interferometric Synthetic Aperture Radar (InSAR) interferograms is crucial for efficient geohazard monitoring, yet it remains fundamentally challenged by severe phase ambiguity and complex coherence noise. While the Segment Anything Model (SAM) offers a powerful foundation for segmentation, its direct transfer to wrapped phase data is hindered by a profound spectral domain shift, which suppresses the high-frequency fringes essential for boundary delineation. To bridge this gap, we propose WILD-SAM, a novel parameter-efficient fine-tuning framework specifically designed to adapt SAM for high-precision landslide detection on wrapped interferograms. Specifically, the architecture integrates a Phase-Aware Mixture-of-Experts (PA-MoE) Adapter into the frozen encoder to align spectral distributions and introduces a Wavelet-Guided Subband Enhancement (WGSE) strategy to generate frequency-aware dense prompts. The PA-MoE Adapter exploits a dynamic routing mechanism across heterogeneous convolutional experts to adaptively aggregate multi-scale spectral-textural priors, effectively aligning the distribution discrepancy between natural images and interferometric phase data. Meanwhile, the WGSE strategy leverages discrete wavelet transforms to explicitly disentangle high-frequency subbands and refine directional phase textures, injecting these structural cues as dense prompts to ensure topological integrity along sharp landslide boundaries. Extensive experiments on the ISSLIDE and ISSLIDE+ benchmarks demonstrate that WILD-SAM achieves state-of-the-art performance, significantly outperforming existing methods in both target completeness and contour fidelity.