2020/04/02 by Despoina Paschalidou, Luc Van Gool, Paschalidou, Despoina +3 · 3 citations
Computer Science · Earth and Planetary Sciences · Engineering · #3D Shape Modeling and Analysis #3D Surveying and Cultural Heritage #Advanced Vision and Imaging #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences
paper · pdf · doi:10.48550/arxiv.2004.01176
openalex publication_date 2020/04/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Humans perceive the 3D world as a set of distinct objects that are\ncharacterized by various low-level (geometry, reflectance) and high-level\n(connectivity, adjacency, symmetry) properties. Recent methods based on\nconvolutional neural networks (CNNs) demonstrated impressive progress in 3D\nreconstruction, even when using a single 2D image as input. However, the\nmajority of these methods focuses on recovering the local 3D geometry of an\nobject without considering its part-based decomposition or relations between\nparts. We address this challenging problem by proposing a novel formulation\nthat allows to jointly recover the geometry of a 3D object as a set of\nprimitives as well as their latent hierarchical structure without part-level\nsupervision. Our model recovers the higher level structural decomposition of\nvarious objects in the form of a binary tree of primitives, where simple parts\nare represented with fewer primitives and more complex parts are modeled with\nmore components. Our experiments on the ShapeNet and D-FAUST datasets\ndemonstrate that considering the organization of parts indeed facilitates\nreasoning about 3D geometry.\n