A new algorithm based on machine learning developed by researchers from Stanford University, allows to identify poor areas using data from satellites.
According to the world Bank, nearly 900 million people worldwide live on less than 1.9. Despite the fact that this is a global problem, identifying the location of impoverished areas on earth is a difficult task that requires a huge amount of money. So researchers from Stanford University have invented a method that allows using the satellite and artificial intelligence to find poor areas.
At the time, as usual the satellite images in themselves do not carry any useful information about the economic level of a particular area, researchers have invented an innovative method. They realized that through the analysis and comparison of daytime and nighttime shots can give concrete expression to the economic prosperity of the district. This helps researchers a special machine learning algorithm, which eventually begins to better define the level of poverty in the photos.
Night shots give more correct information on the prosperous area, because the system can see the light level, which plays an important role in this method. Fluorescent images allow you to look at other aspects of the development of the area, such as the construction of roads and others.
The researchers used a new system for analysis of several regions in Africa. As shown by tests, the system really has great potential.