MAS – Active Satellite Monitoring

MAS – Active Satellite Monitoring is a solution that uses artificial intelligence tools such as convolutional neural networks and computer vision to monitor changes in critical infrastructure.

The MAS use case is based on the continuous surveillance and monitoring of points or areas of interest using Earth observation images obtained via remote sensing, typically from satellites, and their analysis using georeferenced image processing and artificial intelligence techniques.

The images used are those provided by the European Space Agency’s Copernicus programme, as it enables free and periodic access to most of the planet’s land surface. In this case, the images are delivered by a server from an internal tool developed for another project to make satellite images available: the REPOSAT project, funded by RED.ES under the call for Grants for the development of the technological offering in digital content 2020, file number 2020/0820/00102245.

This has made it possible to develop a solution that eliminates the cost of accessing publicly funded data from the outset, limiting the effort to the computational resources required to develop and maintain the tool.

Two different artificial intelligence models have been developed: one based on convolutional neural networks to monitor known and clearly identified anomalies, and a second normative analysis model, using unsupervised learning, for broad-spectrum detection of anomalies not identified a priori. This enables adaptation to potential future changes in the specific characteristics of the structures to be monitored, through the use of transfer learning with the convolutional network model, and through its own design characteristics with the unsupervised vision model.

Financial data