Earthquake-affected building segmentation model

This dataset contains the training, validation, and test data used to develop and evaluate a U-Net deep learning (DL) segmentation model for building detection in earthquake-affected areas from very high-resolution WorldView-2 imagery. It consists of satellite image patches and corresponding binary building masks for pixel-level segmentation, representing diverse post-earthquake urban environments. The dataset supports deep learning-based building extraction and damage-related spatial analysis in disaster-affected regions.