Facebook presented maps of Latin American population density created with satellite images and census data.
Facebook presented high-resolution population maps of Latin American countries, prepared with the census data of each country and with satellite images.
These maps are the most detailed in the world to date: they estimate how many people live for each 30-meter grid, in addition to providing demographic information. For example, through them, you can check the number of children under five years old, as well as that of the population of young people and older adults.
Maps made with AI
The maps created by Facebook are three times more detailed than any other source, thanks to the fact that they combined the potential of Artificial Intelligence with satellite images and public and commercial data sets for their elaboration.
These demographic letters are the product of two years of work between Facebook and Columbia University to use satellite imagery and census data for connectivity projects. After collaborating with groups such as the American Red Cross and their Missing Maps project, they realized that these maps could also be used to improve public health.
Non-profit organizations and researchers can benefit from population map data to address the social, health, and infrastructure challenges in the region, and thus accelerate compliance with the Sustainable Development Goals.
High-res satellite images exist around the world, but the Facebook AI saved hours of work to identify, among millions of square kilometers, images that contained a city or a village. Facebook’s computer vision system examined 9.8 billion individual photos to determine if they included a building. The team found approximately 94 million buildings in a few days.
Population maps help humanitarian work
In other parts of the world, health organizations are already using AI-driven maps for humanitarian aid, as was the case in Malawi and Mozambique.
In Malawi, high-resolution maps helped the American Red Cross and its Missing Maps project identify areas with and without concentrations of people. This maps helped the Red Cross and other teams make more efficient decisions about where to deploy 3,000 health workers to promote measles vaccination.
After Cyclone Idai in Mozambique, a team from the Harvard School of Public Health, Direct Relief, Nethope, and Northwestern University used the high-resolution map of the country to identify areas where cholera might occur and prioritize them in the storage of treatments.
“When I began my career in USAID working on malaria control, I witnessed the fundamental role that accurate data plays in the effectiveness of humanitarian efforts,” says Laura McGorman, Leader of Data Policies for Wellbeing on Facebook. “The exciting thing about projects like these is that they provide an opportunity for our company to contribute through our experience in learning data and machines.”
“Facebook maps ensure that we focus our volunteers’ time and resources where they are needed most, improving the effectiveness of our programs,” says Tyler Radford, executive director of the OpenStreetMap Humanitarian Team, which participates in the Missing Maps project.
Researchers from the Latin American Public Opinion Project (LAPOP) at Vanderbilt University are also interested in using high-resolution Facebook maps to improve the sampling process in their studios. “Often, researchers working in developing countries should build their samples from old census data and available only at a high level of aggregation,” said Noam Lupu, Associate Professor of Political Science and Deputy Director at LAPOP. “These maps will allow us to identify, in census units and in a better way, where the populations are located, increasing the precision of our samples and the representativeness of our studies.”
The latest maps for Latin America can be downloaded for free on the Facebook page on the Humanitarian Data Exchange portal. For more information about Facebook’s work on high-resolution population density maps and other data efforts for wellness, visit the project website here.