The dataset is made of RGB aerial images of flooded regions.
For the training set, these images are coupled with the corresponding 0/1 masks, indicating the flooded segmented region within the entire aerial images.
For the testing dataset, only the aerial images are present, without the corresponding masks.