Application of Deep Learning to the Establishment of Slope Failure Warning Areas

This study uses a deep learning technique to output the slope failure warning areas from numerical elevation data. This is treated as a semantic segmentation task, with UNet as the network model. The experiment results are compared for cases in which the loss function is used and in which other modifications are made to incorporate the idea of disaster countermeasures.