Description:

MLRSNet is a multi-label high spatial resolution remote sensing dataset designed for semantic scene understanding. It offers diverse perspectives of the world captured from satellites, comprising high spatial resolution optical satellite images. The dataset contains 109,161 remote sensing images, meticulously annotated into 46 categories, with each category holding between 1,500 to 3,000 sample images. Each image maintains a fixed size of 256×256 pixels, featuring various pixel resolutions ranging from approximately 10 meters to 0.1 meters.

Download Dataset

Each image in MLRSNet is tagged with several of 60 predefined class labels, with the number of labels per image varying between 1 and 13. This comprehensive labeling enables the dataset to support multiple applications, including multi-label based image classification, multi-label based image retrieval, and image segmentation. MLRSNet thus serves as a valuable resource for advancing research and development in remote sensing and related fields.

This dataset is sourced from Kaggle.