Now, researchers at Stanford have taken an important first step towards creating a reliable map of impoverished regions of the world by teaching computers to recognize and use nighttime lighting (a proxy for economic wealth) to identify low-income areas using satellite imagery of Nigeria, Tanzania, Uganda, Malawi and Rwanda.
The research team published their findings today in Science (subscription required).
“There are few places in the world where we can tell the computer with certainty whether the people living there are rich or poor,” said lead author and School of Engineering doctoral student Neal Jean, in a Stanford press release. “Without being told what to look for, our machine-learning algorithm learned to pick out of the imagery many things that are easily recognizable to humans – things like roads, urban areas and farmland.”
The team's machine-learning algorithm predicted the locations of impoverished areas better than existing approaches, and it used publicly available data.
"It’s cheap and scalable," said co-author Stefano Ermon, PhD. "It could be used to map poverty around the world in a very low-cost way.”
Previously: U.S. Census Bureau releases new data on income, poverty, and health insurance coverage and Bill Gates on improving health care, bolstering education and fighting global poverty
Photo by David Brossard