A Google AI initiative to guide humanitarian efforts
To support applications for social good, Google AI published the Open Buildings dataset. This extensive open dataset includes building footprints, which are detections of buildings on an area of land. These detections are created with high-resolution satellite images and machine learning tools. Currently, the dataset includes 516 million building detections, across an area of 19.4 million km2 on the continent of Africa (64% of the continent).
Knowing whether or not a building is present is extremely useful for:
- Population estimation when there is rapid change, or census information is not up to date
- Coordinating the humanitarian response in case of emergencies and measure the consequences, to natural disasters by estimating the number of buildings or households affected.
- Understanding the human impact on the natural environment, by estimating energy needs and carbon emissions in a certain area, or pressure on protected areas and wildlife due to urbanisation.
- Setting up an addressing system as the absence of a formal addresses limits access to social benefits and economic opportunities.
- Vaccination planning by knowing the density of the population and settlements, demand can be anticipated, and resources can be allocated accordingly.
- Calculating statistical indicators and combining datasets to predict the mean travel time to the nearest hospital for instance.
Curious to learn more? Find more information and explore the dataset and visualisations!