Using AI to identify unmapped schools and increase the accuracy of schools’ geo-location

Go back to programme

Using AI to identify unmapped schools and increase the accuracy of schools’ geo-location

Giga uses AI to identify previously unmapped schools and to increase the accuracy of school’s geo-location. More precise school locations are pivotal to make decisions about how to connect schools to the internet – from selecting appropriate connectivity technologies, to decisions on how to aggregate and pool demand, to identify infrastructure gaps for further investment. Giga uses a blend of convolutional neural networks (ConvNext) and model interpretability methods (GradCAM) on top of high-resolution satellite images in its AI-assisted school mapping efforts. Giga also uses AI to find the lowest cost network configuration to connect schools to the internet, considering different technologies and the existing infrastructure within a country (School Price Calculator).

Share this session

Are you sure you want to remove this speaker?