Back to the blog

An automated satellite image interpretation application based on artificial intelligence is awarded federal R & D funding


Montreal, November 26, 2018 – Effigis Geo-Solutions (Effigis) announced a CANARIE federal funding for a research and development project led by principal investigator Yacine Bouroubi, who is a collaborator at Effigis and professor in the Applied Geomatics Department at Université de Sherbrooke. The project consists in developing a software platform that will automate the interpretation of satellite images to identify land use and ground objects.

So far, image interpretation is done by specialists, often from aerial photographs. Trading places with humans, the artificial intelligence algorithms that will be developed thanks to GeoImageNet platform are expected to shorten the image analysis process from months to hours. “Artificial intelligence is very successful in recognizing objects found in large numbers, such as dogs, houses, cars. This is achieved using neural networks”, Professor Bouroubi said. “This technique is expected to be just as successful with remote sensing, which consists in detecting distant objects in very high-resolution satellite images”, he continued.

Applications of the GeoImageNet Artificial Intelligence Platform

Using deep learning, i.e. by training its network of artificial brain cells, the “brain” of the platform should be capable of generating thematic maps that identify land use and cover. For example, a map could highlight high-density residential areas, and vegetation or soil types.

Thanks to object detection techniques, the platform will also be capable of recognizing objects that can often be seen from the sky. “For instance, one might want to find grain silos in a given area, get the car count of a major city, appraise traffic density on the road network or count the number of trees or pools on private land,” Professor Bouroubi explained.

About the scientist and Effigis collaborator

Professor Yacine Bouroubi is a research and development collaborator at Effigis. His field of interest is the application of artificial intelligence techniques to very high-resolution satellite images. He holds a PhD degree in remote sensing from Université de Montréal and is professor in the Department of Applied Geomatics at the Université de Sherbrooke.

About CANARIE

CANARIE is a non-profit organization primarily funded by the Government of Canada. Its core purpose is the advancement of Canada’s knowledge and innovation infrastructure.

Share
Share on FacebookGoogle+Tweet about this on TwitterShare on LinkedInEmail to someone