A new study presents a deep learning and remote sensing framework that accurately identifies building materials, enabling high-resolution material intensity databases essential for sustainable urban planning.

This scalable technology aids in reducing embodied carbon, enhancing energy efficiency, and promoting urban circularity.

Developed by Peking University and the University of Southern Denmark, the framework uses Google Street View, satellite data, and Convolutional Neural Networks (CNNs) to classify roof and façade materials with high precision.

Advanced visualisation techniques like Grad-CAM improve model transparency.

This innovation supports targeted retrofits, carbon reduction, and sustainable city planning, marking a significant advance in decarbonising the built environment.