Abstract

We propose a new approach for semantic segmentation of 3D city models. Starting from an SfM reconstruction of a street-side scene, we perform classifification and facade splitting purely in 3D, obviating the need for slow imagebased semantic segmentation methods. We show that a properly trained pure-3D approach produces high quality labelings, with signifificant speed benefifits (20x faster) allowing us to analyze entire streets in a matter of minutes. Additionally, if speed is not of the essence, the 3D labeling can be combined with the results of a state-of-the-art 2D classififier, further boosting the performance. Further, we propose a novel facade separation based on semantic nuances between facades. Finally, inspired by the use of architectural principles for 2D facade labeling, we propose new 3D-specifific principles and an effificient optimization scheme based on an integer quadratic programming formulation.