The development of tools for the generation of 3D city models was initiated almost two decades ago, still, research on automatic city modelling remains an active area. Initial work aimed on automatic systems for the efficient 3D reconstruction of polyhedral building objects. Meanwhile, colored triangle meshes are standard data sources for the visualization of urban areas. However, their classification to semantic classes such as buildings, vegetation, impervious surfaces, roads, etc. remains an open research issue. Current work aims on the use of hybrid training data: synthetically generated urban meshes from procedural modelling, as well as real world meshes generated from airborne LiDAR and textured with oblique UAV imagery in order to perform a semantic segmentation for each face in the triangle mesh by means of a Deep Learning architecture tailored for this purpose.
Tutzauer, P. & Haala, N.  Processing of Crawled Urban Imagery for Building Use Classification. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-1/W1, 143-149.
Laupheimer, D., Tutzauer, P., Haala, N. & Spicker, M. (2018) Neural Networks for the Classification of Building-use from Street-View Imagery, ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-2, 177-184, 2018