rslearn

rslearn is a tool for developing remote sensing datasets and models. rslearn helps with:

1. Developing remote sensing datasets, starting with defining spatiotemporal windows (3D boxes in height/width/time) that are roughly equivalent to training examples.
2. Importing raster and vector data from various online or local data sources into the dataset.
3. Fine-tuning remote sensing foundation models on these datasets.
4. Applying models on new locations and times.

New to rslearn?

For a quick introduction to rslearn, walk through the Quickstart. The quickstart builds a dataset with paired Sentinel-2 images and ESA WorldCover land cover labels, and fine-tunes an OlmoEarth model for land cover segmentation.

For a deeper introduction:

• Read Core Concepts, which summarizes key concepts in rslearn, including datasets, windows, layers, and data sources.
• Check out the Usage Overview, which walk through the typical workflow when using rslearn, from defining windows to training models.
• Browse through the Examples and find one that's relevant to your project.

Installation

rslearn requires Python 3.11+ (Python 3.12 is recommended).

git clone https://github.com/allenai/rslearn.git
cd rslearn
pip install .[extra]

Contact

For questions and suggestions, please open an issue on GitHub.