Refocusing Radiant Earth Foundation’s efforts to impact global development with machine learning
(13 May 2019 - Radiant Earth Foundation) Data drives decisions. Whether it’s the number on a scale signaling the need to diet or a satellite image showing the extent of flooding for disaster response, data, imagery, and the resulting analyses they enable guide valuable insights and actions.
Radiant Earth Foundation was founded on the premise that much of the world’s best data and imagery was difficult to find and even more difficult to use because of access issues, making these valuable assets stranded and underutilized.
With a singular focus on global development, the Radiant Earth platform, helped users to discover and unlock these imagery assets and the science of remote sensing to meet the community’s unique needs and challenges whether they be health, climate change, deforestation, or to support innovation around the Sustainable Development Goals.
In the short time that our team pursued this important mission with the development of the Radiant Earth platform, we have witnessed much change in the marketplace:
- Cloud-based data storage and computing is standard practice and an integral part to large-scale remote sensing projects;
- Machine learning is a rapidly evolving field and provides vast opportunities for geospatial problems;
- New Earth observation suppliers and services appear almost daily;
- Existing commercial suppliers are expanding their services and offerings; and
- Numerous platforms now exist with data richness and functional capabilities well beyond our current offering.
Just as imagery enables analysis of the changing environment, these marketplace changes represent key data points that we at the Radiant Earth Foundation must assess to ensure we are best responding to the global development community’s needs and challenges, making the best use of philanthropic dollars and pro-actively positioning our work to impact the future.
After extensive analysis and business planning, we have decided that going forward, our efforts will focus on the development of open training libraries, machine learning models and technology standards that support the use of Earth observation data. As a result, further platform development and operations will cease. Additionally, we will continue to track the remote sensing market and share our observations around the best use of Earth observation data in support of the global development community’s most pressing problems.
This enhanced machine learning focus will enable Radiant Earth Foundation to leverage investments in this field by governments, universities, and the commercial sector, and expand the availability of accurate and diverse training data — the most fundamental requirement to develop machine learning models on Earth observation data.
The centerpiece of our work will be MLHub Earth, which launched in June of 2018. Specifically, we will expand these important commons for training data, models, and standards for different applications. Designed for and by the global development community, MLHub Earth encourages an open, collaborative community and demonstrates the principles that have and will continue to guide the work of the Radiant Earth Foundation.