How satellite images and AI could help fight spatial apartheid in South Africa  

The older Sefala grew to become, the extra she peppered her father with questions concerning the seen racial segregation of their neighborhood: “Why is it like this?”

Now, at 28, she helps do one thing about it. Alongside pc scientists Nyalleng Moorosi and Timnit Gebru on the nonprofit Distributed AI Analysis Institute (DAIR), which Gebru arrange in 2021, she is deploying pc imaginative and prescient instruments and satellite tv for pc photos to investigate the impacts of racial segregation in housing, with the final word hope that their work will assist to reverse it.

“We nonetheless see beforehand marginalized communities’ lives not enhancing,” says Sefala. Although she was by no means alive in the course of the apartheid regime, she has nonetheless been affected by its terrible enduring legacy: “It’s simply very unequal, very irritating.”

In South Africa, the federal government census categorizes each wealthier suburbs and townships, a creation of apartheid and usually populated by Black individuals, as “formal residential neighborhoods.” That census is used to allocate public spending, and when they’re lumped along with richer areas, townships are in impact hidden, disproportionately excluding the individuals dwelling there from entry to assets equivalent to well being providers, schooling facilities, and inexperienced areas. This problem is usually often known as spatial apartheid. 

Raesetje Sefala is deploying satellite tv for pc photos and AI to map out spatial apartheid in South Africa.


Sefala and her crew have spent the final three years constructing an information set that maps out townships as a way to examine how neighborhoods are altering when it comes to inhabitants and dimension. The hope is that it might assist them see whether or not or not individuals’s lives in townships have improved for the reason that authorized dissolution of apartheid.

They did it by accumulating thousands and thousands of satellite tv for pc photos of all 9 provinces in South Africa, and geospatial knowledge from the federal government that reveals the situation of various neighborhoods and buildings throughout the nation. Then they used all this knowledge to coach machine-learning fashions and construct an AI system that may label particular areas as rich, non-wealthy, non-residential, or vacant land. 

In 2021, they found that over 70% of South African land is vacant, they usually noticed how a lot much less land is allotted to townships than to suburbs. It was a affirmation of the inequalities that they had anticipated to see, however the staggering amount of vacant land nonetheless took them aback, says Sefala.

Now they’re sharing the information set with researchers and public service establishments, together with nonprofits and civic organizations working to determine land that could possibly be used for public providers and housing. DAIR plans to make the information let out and accessible on its web site from February 2.

Leave a Comment