Convolutional neural networks and deep learning models have recently been investigated, making it possible to quickly and accurately map landslides, but they haven't been used for multi-temporal landslide mapping in the Himalayas yet. A small landslide inventory across a small region was used for training the earlier models' supervised learning methodology, which was then applied to predict landslides in the area. We suggest a new technique that uses geographically distinct training samples to develop a common methodology that can be applied to develop multi-temporal landslide inventories. In the study region of the Rasuwa district of Nepal, MT landslide inventories are created using RapidEye pictures with a spatial resolution of 5 meters.
Please cite our work if you use our codes: https://doi.org/10.1080/15481603.2023.2182057