Inferring roof semantics for more accurate solar potential assessment

Irène Apra, Carolin Bachert, Camilo Cáceres, Özge Tufan, Ondrej Veselý

In order to determine the available roof area for a more accurate solar potential analysis, this project aims to automatically detect roof obstacles. For this, three different methods were developed and finally merged into one result. The first one relies on a geometry-based classification of AHN3 point clouds and LoD 2.2 building models. The second ones uses an unsupervised image classification of aerial images. Finally, supervised image classification method makes use of the aerial images as well as the BAG footprints and a dataset of manually labelled solar panel polygons. By merging the methods the overall accuracy could be improved. The end result is a LoD2.2 building model in CityJSON format, enhanced with three new attributes per building: the obstacle area on the roof, the available area for installing solar panels, and a Boolean value showing whether the building has existing solar panels or not.

/* */