Harnessing the Power of Earth Foundation Models

Welcome to the IGARSS25 EarthFM tutorial 🥳! This Jupyter book 📖 contains notebooks on how to apply 🧱 Foundation Models to 🛰️ Earth Observation data.

📚 Overview of tutorials¶

1. Introduction to Earth Foundation Models
2. Choosing a Foundation Model
3. Applying Foundation Models to Downstream Tasks

Each tutorial is rendered on this website for easy viewing 👀, but some of them are Jupyter notebooks designed to be ran interactively 💫. See the instructions below on how you can start running the tutorials in no time! 🚀

🌠 Setting up your environment¶

To run these notebooks in an interactive Jupyter session online, 🖱️ click on the button below to launch on Google Colaboratory.

!pip install uv

!git clone https://github.com/developmentseed/igarss25tutorial.git
!cd igarss25tutorial && git switch eofm_eval_notebook

!cd igarss25tutorial && uv export -o pylock.toml > pylock.toml
# ! && uv pip sync --preview --system pylock.toml

!cd igarss25tutorial && pip install -r pylock.toml

Alternatively, you can choose to run the Jupyter notebooks on another cloud provider with GPU instances such as Sagemaker Studio Lab.

💻 Creating a local environment for running tutorials¶

If you prefer to run the 🧑‍🏫 tutorials with a local installation instead, then follow along! For this IGARSS25 workshop, we recommend creating a virtual environment with uv and installing the 🐍 Python libraries inside.

Here’s the instructions to install the igarss25tutorial environment:

1. Ensure that you have the uv package manager installed.

2. Make a folder called ‘igarss25tutorial’. This will be where you will put all the Jupyter notebooks and data files 🗃️ used in the workshop.

3. Download a copy of the ‘pyproject.toml’ and ‘uv.lock’ files which contains a 📄 list of dependencies required to run the tutorials in this workshop. Get it at pyproject.toml and uv.lock

4. Run the following commands on the 🧑‍💻 command-line to create the virtual environment

   cd /path/to/igarss25tutorial
   uv sync --locked

5. Once the installation is completed 🏁, launch Jupyter Lab as follows:

   source .venv/bin/activate
   uv run --with jupyter jupyter lab

   This should open up a page in your default browser. If not, you can click and open the 🔗 link that says http://localhost:8888/lab?token=... in your command-line terminal and this will take you to the Jupyter Lab page.

6. Download the Jupyter notebook(s) you want to run (e.g. https://www.developmentseed.org/igarss25tutorial/tut3_placeholder) using either the download button on the ↗️ top right (select ‘.ipynb’) or from GitHub at https://github.com/developmentseed/igarss25tutorial. Make sure to put the *.ipynb file(s) inside of the ‘igarss25tutorial’ folder.

7. Open the Jupyter notebook in the left-pane file browser, e.g. by 🖱️ double-clicking on tut3_placeholder.ipynb. You are now ready to run through the course materials 🎉!