We propose a new approach for semantic segmentation of 3D city models. Starting from an SfM reconstruction of a street-side scene, we perform classification and facade splitting purely in 3D, obviating the need for slow image-based semantic segmentation methods. We show that a properly trained pure-3D approach produces high quality labelings, with significant speed benefits (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 classifier, 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-specific principles and an efficient optimization scheme based on an integer quadratic programming formulation.

Improvement in acquisition systems, has resulted in the ability to capture more realistic 3D models of real world objects, creating a need for better data processing techniques, as in the case of text and images. In this paper, we address the issue of learning class-specific, deformable, 3D part-based structure for object part localization in 3D models/scenes. We employ an inference framework upon fully connected part-based graphs inspired by Pictorial Structures (PS), which combine the local appearance of parts and the long-range structural properties. Using efficient tools for learning the model and performing inference, we show good results on a variety of classes, outperforming PS [7] and ISM [15]. Further, a similar inference framework is employed to find dense correspondences between 3D models, seeded by the above object part localization. Our results show promise for application in more complex 3D processing tasks such as part retrieval, pose estimation, scene understanding and recognition. © 2011. The copyright of this document resides with its authors.

This paper presents a new efficient technique for large-scale structure from motion from unordered data sets. We avoid costly computation of all pairwise matches and geometries by sampling pairs of images using the pairwise similarity scores based on the detected occurrences of visual words leading to a significant speedup. Furthermore, atomic 3D models reconstructed from camera triplets are used as the seeds which form the final large-scale 3D model when merged together. Using three views instead of two allows us to reveal most of the outliers of pairwise geometries at an early stage of the process hindering them from derogating the quality of the resulting 3D structure at later stages. The accuracy of the proposed technique is shown on a set of 64 images where the result of the exhaustive technique is known. Scalability is demonstrated on a landmark reconstruction from hundreds of images.