The TerraTorch and TerraKit libraries have joined the TorchGeo Organization! https://github.com/torchgeo is now home to 3 of the most influential and widely used open-source GeoAI libraries:
- TorchGeo: the core of the geospatial deep learning ecosystem, with more geospatial datasets and pre-trained foundation models than all other GeoAI libraries combined
- TerraTorch: higher-level abstractions that build on top of TorchGeo to make foundation model fine-tuning simple and intuitive for users from all backgrounds
- TerraKit: provides connectors and interfaces for NASA, ESA, and other data stores, allowing users to effortlessly download and build AI-ready datasets
To ensure even greater collaboration and long-term maintenance of these libraries, we are also excited to announce 2 new members joining our Technical Steering Committee:
- @Romeo Kienzler: AI Research Engineer at IBM Research, co-creator and maintainer of TerraTorch and TerraKit, and co-author of Prithvi WxC
- @Isabelle Wittmann: Research Software Engineer at IBM Research, maintainer of TerraTorch, co-organizer of the Embed2Scale EarthVision challenge, and Earth2Vec, TerraCodec, and NeuCo-Bench co-author
Romeo and Isabelle’s expertise in foundation models, neural compression, and embeddings will further strengthen the TorchGeo ecosystem, and ensure greater collaboration and compatibility between all projects we govern.
