Rural settlement Planet remote sensing image datase

This dataset is a deep learning training dataset used for remote sensing automatic extraction of rural settlements. The data core consists of high-resolution satellite images from Planet and their corresponding manually interpreted annotation masks. This dataset is specifically designed for training and evaluating semantic segmentation models for remote sensing images, with the aim of automatically and accurately extracting settlement spatial ranges from complex rural backgrounds. The task challenges include dealing with the diversity of settlement forms, spectral confusion with background features, and small target detection. The dataset is organized in the form of "image mask" pairing, usually stored in separate newjpg and newmask directories, ensuring strict file name correspondence for easy programmatic reading and batch loading.