Figure 7
From: A new strategy to map landslides with a generalized convolutional neural network
Pre-processing workflow adopted for training of CNN. (A) Systematic tiling of input images with 50% overlap. In this study , a tile size of \(224 \times 224\) was used as an input to the CNN. The tile size in display is for representation purposes and does not scale to the actual base resolution used in training the CNN. (B) Steps involved in the data-augmentation of input images with their corresponding landslide mask.
