EuroFab research project

The “EuroFab: European Urban Fabric Classification using Artificial Intelligence” project brings an innovative perspective to understanding the structure of human settlements by capturing their physical composition using explainable data science in combination with Earth Observation (EO), and through state-of-the-art AI, uncovering their temporal dimension.

Specifically, the solutions develop a two-step system combining mathematical characterisation of built-up fabric using techniques known as urban morphometrics, providing a rich hierarchical classification backed by rich interpretable numerical descriptions, and state-of-the-art neural networks to recognise such classification from open optical satellite imagery (Sentinel 2), allowing both the detection of individual types in areas with ground data of lesser quality and the unfolding of such classification over time, providing a unique temporal look at how human settlements evolve.

The output paves the way to map urban development in the European continent over both space and time, covering every building from the smallest hamlet to the largest metropolis.

EuroFab helps us understand the habitats we built and efficiently affect policy at European, national and local levels ensuring the future of “urban” is sustainable, resilient and data driven.

The proposed project is an initiation of a more ambitious undertaking. It focuses on developing scientific and technical solutions based on novel AI methods, in conjunction with strategically shaping the outputs and framework to serve various stakeholders directly involved in policy, planning, and research.

For more details, please refer to individual Technical Notes.

The activities reported in this document were carried under a programme of, and funded by, the European Space Agency. View expressed in the publication can in no way be taken to reflect the official opinion of the European Space Agency.