This dataset contains images of flood-affected areas and corresponding mask images showing the water areas.

The dataset contains 290 images and self-annotated masks. The mask images were labeled using the open source data labeling software Label Studio The task is to create a segmentation model that can accurately segment the water area in a given flood-affected area image.

Such models can be used for flood investigations, better decision making, and planning. Due to less data, pre-trained models and data augmentation can be used.

Naviagate Dataset:

1. image: The folder containing all flood images.
2. Mask: The folder containing all mask images.
3. metadata.csv: csv file mapping image names to their masks.