AI helps aid reach uncharted regions of the developing world

Artificial intelligence is being used to help map uncharted streets in developing countries to allow governments and aid groups to determine where people live and how to reach them.

Large numbers of homes in Africa, Asia and Latin America are not currently represented on any maps, meaning they can be left out of services, planning, decision-making and politics.

But a new Google data set of more than 1.8bn buildings across Africa, Asia, Latin America and the Caribbean has used artificial intelligence to turn blurry satellite photos into detailed maps outlining individual buildings.

The latest Open Buildings dataset also contains maps drawn up from 2016 to 2023, showing how towns and cities have expanded and changed in some of the fastest-growing countries in the world.

The data has already been used on projects such as targeting children missed from vaccination campaigns.

Prof Andrew Tatem, director of the WorldPop team at Southampton University, who has been using the data to help produce new population estimates, said: “Understanding where people live is vital for making sure that resources are distributed fairly and that no one is left behind in delivering services like healthcare.”

“Google’s Open Buildings dataset has been a fantastic addition to the open data in our field that has supported more accurate population mapping, and the new temporal dataset unlocks opportunities to better capture the rapid demographic changes we continue to see globally.”

In one example, the WorldPop team worked with the African Field Epidemiology Network to understand how many children had not been vaccinated in a region of Katsina State in northern Nigeria.

There were no good maps of the area, so they used the data to plan and execute a door-to-door campaign to follow up with families and make sure that children in the area received routine vaccinations.

Google scientists said they had used poorer quality satellite images, because this was often all that was available in some parts of the world.

However they admitted that meant the maps may have mistakes and limitations.

The AI model relies on cloud-free images to get the best results and these can be difficult in some cloudy climates. Tiny buildings, like informal shacks can also be missed.

A statement from Google said: “Maps are a lifeline to many things we need.

“For people to receive essential services, like electricity and running water, and to be accounted for in crisis response, decision-makers need to first know where they are.

“By creating maps, we can help decision-makers understand the current environment and ensure that everyone is reached.”

Protect yourself and your family by learning more about Global Health Security