The Dockerfile has been modified to copy the package from local files and install in an editable configuration rather than installing from the public release. This takes longer on startup, as Terratorch must be installed from local files each time the container starts for edits to be reflected in those files. To disable this behavior and build a more permanent version of the Docker image from local files, swap out the bash command at the end of the Dockerfile for the copy command in the middle of the Dockerfile, and uncomment Terratorch in the pip install line.
The compose.yml is the preferred way of running this workflow. A template has been provided.
The Embedding Generation task has been added to the Terratorch package
The embedding generation workflow relies on the terratorch predict method. Connect to the terminal of the container and activate the Terratorch virtual environment using:
source /opt/app-root/src/venv/bin/activate
Using this line from the official tutorial as an example:
terratorch predict -c burn_scars_config.yaml --predict_output_dir outputs/ --data.init_args.predict_data_root examples/ --ckpt_path Prithvi_EO_V2_300M_BurnScars.pt
e.g.
terratorch predict -c /opt/app-root/src/terratorch/examples/confs/embedding_generation/burn_scars_embedding_config.yaml --data.init_args.predict_data_root /data/Prithvi-EO-2.0-300M-BurnScars/examples/
Change the output dir and ckpt path according to the local filestructure.
TerraTorch is a PyTorch domain library based on PyTorch Lightning and the TorchGeo domain library for geospatial data.
Watch the latest recording on YouTube: Earth observation foundation models with Prithvi-EO-2.0 and TerraTorch
TerraTorch’s main purpose is to provide a flexible fine-tuning framework for Geospatial Foundation Models, which can be interacted with at different abstraction levels. The library provides:
- Convenient modelling tools:
- Flexible trainers for Image Segmentation, Classification and Pixel Wise Regression fine-tuning tasks
- Model factories that allow to easily combine backbones and decoders for different tasks
- Ready-to-go datasets and datamodules that require only to point to your data with no need of creating new custom classes
- Launching of fine-tuning tasks through CLI and flexible configuration files, or via jupyter notebooks
- Easy access to:
- Open source pre-trained Geospatial Foundation Model backbones:
- Backbones available in the timm (Pytorch image models)
- Decoders available in SMP (Pytorch Segmentation models with pre-training backbones) and mmsegmentation packages
- Fine-tuned models such as granite-geospatial-biomass
- All GEO-Bench datasets and datamodules
- All TorchGeo datasets and datamodules
In order to use th file pyproject.toml it is necessary to guarantee pip>=21.8. If necessary upgrade pip using python -m pip install --upgrade pip.
For a stable point-release, use pip install terratorch==<version>.
To get the most recent version of the main branch, install the library with pip install git+https://github.com/IBM/terratorch.git.
TerraTorch requires gdal to be installed, which can be quite a complex process. If you don't have GDAL set up on your system, we recommend using a conda environment and installing it with conda install -c conda-forge gdal.
To install as a developer (e.g. to extend the library):
git clone https://github.com/IBM/terratorch.git
cd terratorch
pip install -r requirements_test.txt
conda install -c conda-forge gdal
pip install -e .
To install terratorch with partial (work in development) support for Weather Foundation Models, pip install -e .[wxc], which currently works just for Python >= 3.11.
To get started, check out the quick start guide.
Developers, check out the architecture overview.
TerraTorch: The Geospatial Foundation Models Toolkit on arXiv
This project welcomes contributions and suggestions. Ways to contribute or get involved:
- Join our Slack
- Create an Issue (for bugs or feature requests)
- Contribute via PR
- Join our duoweekly community calls taking place Tuesdays 4:30 PM - 5 PM CEST and Thursdays 2:30 PM - 3 PM CEST.
You can find more detailed contribution guidelines here.
A simple hint for any contributor. If you want to meet the GitHub DCO checks, just do your commits as below:
git commit -s -m <message>
It will sign the commit with your ID and the check will be met.
This project is primarily licensed under the Apache License 2.0.
However, some files contain code licensed under the MIT License. These files are explicitly listed in MIT_FILES.txt.
By contributing to this repository, you agree that your contributions will be licensed under the Apache 2.0 License unless otherwise stated.
For more details, see the LICENSE file.