Human activity analysis

Extracting actionable analytics on remote area development from satellite imagery with a custom computer vision solution


An insurance company needed to monitor human activity in inaccessible areas based on historic and current satellite images. They also wanted to compare different areas, analyzing differences in current data, as well as changes over time, to assess which areas were developing the fastest.


Cognition implemented SAHARA (Satellite Human Activity Recognition and Analysis), a custom AI solution to detect and measure human activity, combining advanced computer vision technology with data analysis.

Cognition used satellite images provided by the client to calibrate the computer vision technology. Together with the client, they defined which parameters to deliver as part of the data analysis. Cognition’s team adapted the report format and the user interface of the SAHARA tool to the client’s needs.

Typically an area needs several satellite images to be adequately covered. SAHARA allowed users to apply self-selected images, or choose which areas to analyze within an image via a graphical interface.


The system was able to deliver current information about human activity in any area of the Earth, assess its evolution over time, and compare different areas with no need to review (often inaccurate) official reports or send a team to assess the human activity in person. The solution saved significant time and greatly reduced the cost and risk associated with collecting the information onsite.