Coastline extraction from ALOS-2 satellite SAR images

The continuous monitoring of a shore plays an essential role in designing strategies for protection against erosion. To avoid the effect clouds and sunlight, satellite-based imagery with synthetic aperture radar is used to provide required data. In contrast standard model-driven methods, we present deep-learning-based approach detect coastlines such We split process into data preprocessing, model training, inference, ensembling, postprocessing describe best techniques each parts. deal small training dataset, propose novel multi-sample mosaicing augmentation that helps deep neural network models reduce overfitting during training. Our solution has been validated real Global Positioning System location worldwide competition organized by Signate Japan Aerospace Exploration Agency, where it was runner-up among 109 teams from whole world.