A study on the identification of rocks in non-obvious landslides based on instance segmentation

Accurate identification of subsurface structures beneath non-obvious landslides is crucial for the prevention and management of such geological disasters. To detect potential rock anomalies beneath landslides, this study proposes an improved Mask R-CNN instance segmentation network that integrates ConvNeXt and Hybrid Task Cascade. The results show that the proposed method achieves high accuracy in identifying and segmenting overlapping subsurface targets in ground-penetrating radar.