The R+D+I area of the public company of the Government of Navarre shared, at the world’s leading event on Earth Observation ― the International Symposium on Geosciences and Remote Sensing, organized by the IEEE ― the first work published to date that applies diffusion technology to improve the detection of buildings in aerial images

Tracasa Global, a public company of the Government of Navarre, recently presented its innovative advances in Artificial Intelligence and Remote Sensing in the United States; specifically, at the world’s premier event on Earth Observation today: the International Symposium on Geosciences and Remote Sensing (IGARSS 2023), organized by the IEEE (Institute of Electrical and Electronics Engineers), held this year in Pasadena (California).
The R+D+I area shared, with some 4,000 experts from around the world, its work entitled “Diffusion models for the semantic segmentation of remote sensing images”, which is the first one published to date that applies diffusion technology to improve the detection of buildings in aerial images.
How do diffusion models work?
Diffusion technology is a form of Artificial Intelligence that drives tools capable of creating content based on a specific indication; usually, in text form. These diffusion models are trained using massive data sets and, through an iterative process, progressively learn to generate images that are as close as possible to the request made.
In the case of Tracasa Global and Tracasa Instrumental, the work managed to demonstrate, for the first time, that diffusion models, to date mostly articulated based on indications in the form of text, are also capable of extracting buildings from a remote sensing image, allowing for progress in the semantic segmentation of Earth Observation images. Therefore, it was presented as the first work that applies diffusion technology to the detection of buildings.
An international leader in remote sensing
Tracasa Global and Tracasa Instrumental’s contributions at top-level international events in the field of Artificial Intelligence and Remote Sensing have become a constant in recent years. The public company of the Government of Navarre has claimed victories and top positions at several international Artificial Intelligence and Earth Observation contests.
In fact, last year, at the competition entitled AI4FoodSecurity, organized by the European Space Agency, Radiant Earth Foundation, TUM-DLR and Planet, Tracasa Global and Tracasa Instrumental’s R+D+I area emerged victorious in a competition whose objective was to classify types of crops with Artificial Intelligence techniques based on time series of images from the Sentinel-1, Sentinel-2 and Planet Fusion satellites.