GeoPhysAdapter: Scale-Matched Geophysical Adaptation for Cross-Domain Landslide Mapping with Vision Foundation Models
2026-08-10 • Artificial Intelligence
Artificial IntelligenceComputer Vision and Pattern Recognition
AI summaryⓘ
The authors studied how to improve mapping of newly formed landslides, which usually lack clear labels for training models. They used a frozen vision foundation model and combined it with local terrain, material, and rainfall information at two levels: pixels and whole landslide candidates. Their method, GeoPhysAdapter, better reduces false alarms by using this physical context, especially when applied to entire landslide shapes rather than individual pixels. Testing on a large, diverse dataset showed that adapting at the landslide body level cut errors much more than pixel-level adjustments. Their work helps make landslide detection more reliable across different regions and events.
landslide mappingvision foundation modelcross-domain transferfalse positivesterrain datarainfall triggeringuncertain geographic context problempixel-level adaptationcandidate landslide bodyintersection over union (IoU)
Authors
Zhihang Liu, Mei-Po Kwan, Jinlin Wu, Hao Li
Abstract
Newly triggered landslides rarely carry immediate annotations, so cross-domain transferability determines the value of landslide mapping for emergency response and regional risk assessment. Vision foundation models have strengthened representational transfer, yet on unseen regions, events, and data sources they still generate high-confidence false alarms. Terrain, material, and rainfall triggering can constrain such errors, but their supports are local, regional, and event-scale, so that resampling onto a 10~m grid misaligns them with the segmentation decision unit and compounds the uncertain geographic context problem (UGCoP). We propose GeoPhysAdapter, which anchors on a frozen vision foundation model, restricts terrain, material, and triggering to dense spatial guidance, regional modulation, and event-timing forcing, and applies bounded adaptation at two decision units, the pixel and the candidate landslide body, reverting exactly to the visual prediction where support is insufficient. On an event-isolated PILD dataset of four public sources, 55 global landslide events, and 7,890 test samples, 70.3% of cross-domain false-positive mass lies in near-pure spurious bodies of median equivalent diameter 207m, matching coarse-prior support rather than the pixel. Pixel-level adaptation removes a net 507,817 erroneous pixels and reduces error by 7.76%, whereas raising the decision unit to the candidate body, under identical samples, anchor, and baseline, increases error reduction to 23.99%, approximately 3.1 times the pixel-level effect, improves IoU by 0.031 (14.2% relative), and corrects 9.92 pixels per pixel harmed. The data and code are publicly available at: https://github.com/Liu-Zhihang/geophysadapter.