Facebook has already created a population map based on AI. However, the site has recently provided very detailed maps of population density covering the “majority” of Africa, helping aid agencies know where to go when providing assistance. But changes to the formula have kept high precision even across Africa, which has an area of 16 million square miles.
According to the US site Engadget, the project was based on a neural network that classifies 11.5 billion satellite images with a huge area (64 x 64 pixels each), which can determine where people live based on the presence of buildings in each frame, Combined with census data and some verification, Facebook can identify 110 million home sites within days, bypassing areas they know are uninhabited.
It is not a complete picture of the continent, Facebook has some ways to go before it can identify the land, more countries are coming, and much of a continent is coming. Facebook is trying to succeed through its population maps in countries like Malawi Tanzania, where he helped provide vaccines and electricity to rural areas, while new data could amplify those efforts and ensure that remote communities receive the assistance they need.