Sunkel, MartinJansen, SilkeWand, MichaelSeidel, Hans-PeterI. Navazo, P. Poulin2015-02-282015-02-2820131467-8659https://doi.org/10.1111/cgf.12040This paper addresses the problem of detecting objects in 3D scans according to object classes learned from sparse user annotation. We model objects belonging to a class by a set of fully correlated parts, encoding dependencies between local shapes of different parts as well as their relative spatial arrangement. For an efficient and comprehensive retrieval of instances belonging to a class of interest, we introduce a new approximate inference scheme and a corresponding planning procedure. We extend our technique to hierarchical composite structures, reducing training effort and modeling spatial relations between detected instances. We evaluate our method on a number of real-world 3D scans and demonstrate its benefits as well as the performance of the new inference algorithm.Computer Graphics [I.3.5]Computational Geometry and Object ModelingObject hierarchiesImage Processing and Computer Vision [I.4.8]Scene AnalysisObject recognitionArtificial Intelligence [I.2.10]Vision and Scene UnderstandingShape