LMHLD: A Large-scale Multi-source High-resolution Landslide Dataset for Landslide Detection based on Deep Learning

LMHLD collects remote sensing images of five satellite sensors in seven areas of the world: Wenchuan, China (2008); Rio de Janeiro, Brazil (2011); Gorkha, Nepal (2015); Jiuzhaigou, China (2015); Taiwan, China (2018); Hokkaido, Japan (2018); Emilia-Romagna, Italy (2023), a total of 25,365 patches by setting different patch sizes according to different landslide scale.

Specifically, patch sizes vary across different areas: 32 for image segmentation in Emilia-Romagna, Italy and Gorkha, Nepal; 64 in Rio de Janeiro, Brazil; 128 in Jiuzhaigou, China and Hokkaido, Japan; and 224 in Wenchuan and Taiwan, China. All the above patches constitute LMHLD, a large-scale multi-source high-resolution heterogeneous landslide dataset.

LMHLD.rar includes all the Train, Validation, Test data used for the experiments.

If you use our data, please cite our work published in IEEE Transactions on Geoscience and Remote Sensing. Or if you have any other questions about LMHLD, please contact us: [e-mail].

Paper: Liu G, Wang Y, Chen X, et al. LMHLD: A Large-scale Multi-Source High-Resolution Landslide Dataset for Landslide Detection based on Deep Learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025. (DOI: 10.1109/TGRS.2025.3619062)