veseln/norsc19-eo-learn-workshop

Bridging Earth Observation data and Machine Learning in Python at NoRSC19 conference

Bridging Earth Observation data and Machine Learning in Python

Abstract

Extracting valuable information from satellite imagery is challenging primarily due to large amounts of data. On top of that there is a big lack of techniques allowing for automatic detection and extraction of complex patterns in such spatiotemporal data. Join us to see how we leverage eo-learn to obtain meaningful information from satellite data with just a few lines of Python code.

Description

The availability of open Earth observation (EO) data through the Copernicus and Landsat programs, as well as plethora of commercially available satellite imagery, represents an unprecedented resource for many EO applications, ranging from ocean and land use/land cover monitoring to disaster control, emergency services and humanitarian relief. Large amounts of such spatiotemporal data call for tools that are able to automatically extract complex patterns embedded inside.

eo-learn is a collection of open source Python packages that have been developed to seamlessly access and process spatio-temporal satellite imagery in a timely and automatic manner. It makes extraction of valuable information from satellite imagery as easy as defining a sequence of operations to be performed on satellite imagery. It also encourages collaboration --- the tasks and workflows can be shared, thus allowing for community-driven ways to exploit EO data.

The eo-learn library acts as a bridge between the Earth Observation (EO)/Remote Sensing (RS) field and the Python ecosystem for data science and machine learning. It lowers the entry barrier to the field of RS for non-experts and simultaneously brings the state-of-the-art tools for computer vision, machine learning, and deep learning existing in Python ecosystem to remote sensing experts.

During the workshop we will introduce the eo-learn framework, show examples of tasks dealing with retrieving the EO data (e.g. Sentinel-2, Sentinel-1, DEM), processing it, adding non-EO data (e.g. labels) to the dataset etc. and finally build the whole pipeline to run such workflow for larger areas, thus preparing the data for ML algorithms.

Installation notes

Follow the workshop with just your browser?
You can use the "launch binder" link above at the top of this README, which will launch a notebook instance on Binder with all required libraries installed.

Work on your own computer?
In case you have some time on your hands, and would like to work through the exercises on your machine, the minimal requirements are

• eo-learn (please see installation instructions)
• jupyter

At the moment the recommended way is using a virtual environment (venv, i.e. python3.6 -m venv eo-learn-workshop) or pipenv Installing with conda might prove problematic. On Linux it is recommended to install system packages from CI build instructions first.

Alternatively, create a new environment for this tutorial using the provided environment.yml file:

conda env create --name eo-learn-workshop --file environment.yml
conda activate eo-learn-workshop

Downloading the tutorial materials

Note: We'd like to make this repository a "living" tutorial, so we will be updating the materials with the new versions and new functionalities of the eo-learn package. To update your local copy, you can download the latest version again, or do a git pull if you are using git.

If you have git installed, you can get the tutorial materials by cloning this repo:

git clone https://github.com/sentinel-hub/norsc19-eo-learn-workshop.git

Otherwise, you can download the repository as a .zip file by heading over to the GitHub repository (https://github.com/sentinel-hub/norsc19-eo-learn-workshop) in your browser and click the green "Download" button in the upper right:

Authors

EO Research team at Sinergise ([email protected]).