$17,000 Agriculture Satellite Data Machine Learning Challenge

The agricultural sector makes a substantial contribution to GDP and livelihoods across the developing world. However, regular and reliable agricultural data remains difficult and expensive to collect on the ground.

As a result, policy-makers usually don’t have access to updated data for implementing policies or supporting farmers.

Earth observation satellites provide a wealth of time-series multispectral image data that can be used for developing agricultural monitoring tools. These machine learning solutions support farmers and policy-makers across Africa and the developing world.

The Radiant Earth Spot the Crop Challenge is for you to use time-series of Sentinel-1 and Sentinel-2 multi-spectral data to classify crops in the Western Cape of South Africa.

The machine learning model needs to predict crop type classes for the test dataset using training data generated from ground reference data.

This competition is organized into two parallel tracks:

• One data set: Using time-series of Sentinel-2 multi-spectral data
• Both data sets: Using time-series of both Sentinel-2 and Sentinel-1 (synthetic aperture radar) data.

Winners will release their top solutions under an open-source licence for ongoing use and learning to be eligible for cash prizes. The winners’ solutions will be shared in a well-organized code repository with proper documentation and added to the Radiant MLHub model repository.