Supervised Fitting of Geometric Primitives to 3D Point Clouds

Abstract

Fitting geometric primitives to 3D point cloud data bridges a gap between low-level digitized 3D data and high-level structural information on the underlying 3D shapes. As such, it enables many downstream applications in 3D data processing. For a long time, RANSAC-based methods have been the gold standard for such primitive fitting problems, but they require careful per-input parameter tuning and thus do not scale well for large datasets with diverse shapes. In this work, we introduce Supervised Primitive Fitting Network (SPFN), an end-to-end neural network that can robustly detect a varying number of primitives at different scales without any user control. The network is supervised using ground truth primitive surfaces and primitive membership for the input points. Instead of directly predicting the primitives, our architecture first predicts per-point properties and then uses a differential model estimation module to compute the primitive type and parameters. We evaluate our approach on a novel benchmark of ANSI 3D mechanical component models and demonstrate a significant improvement over both the state-of-the-art RANSAC-based methods and the direct neural prediction.

arXiv commentCVPR 2019 (Oral)

References in corpus (5)

• PCPNET: Learning Local Shape Properties from Raw Point Clouds
• Deep Part Induction from Articulated Object Pairs
• Training Deep Networks with Structured Layers by Matrix Backpropagation
• 3D-PRNN: Generating Shape Primitives with Recurrent Neural Networks
• Differentiation of the Cholesky decomposition

Cited by in corpus (28)

• Linking Points With Labels in 3D: A Review of Point Cloud Semantic Segmentation
• Approximate Convex Decomposition for 3D Meshes with Collision-Aware Concavity and Tree Search
• Learning Part Boundaries from 3D Point Clouds
• PIE-NET: Parametric Inference of Point Cloud Edges
• RFEPS: Reconstructing Feature-line Equipped Polygonal Surface
• Recognising geometric primitives in 3D point clouds of mechanical CAD objects
• PolyGNN: Polyhedron-based Graph Neural Network for 3D Building Reconstruction from Point Clouds
• Reconstructing editable prismatic CAD from rounded voxel models
• SketchGen: Generating Constrained CAD Sketches
• Fitting and recognition of geometric primitives in segmented 3D point clouds using a localized voting procedure
• Learning elementary structures for 3D shape generation and matching
• PrimiTect: Fast Continuous Hough Voting for Primitive Detection
• Augmenting Implicit Neural Shape Representations with Explicit Deformation Fields
• Holistic Parameteric Reconstruction of Building Models from Point Clouds
• BPNet: Bézier Primitive Segmentation on 3D Point Clouds
• Learning Structural Graph Layouts and 3D Shapes for Long Span Bridges 3D Reconstruction
• SALD: Sign Agnostic Learning with Derivatives
• DeformSyncNet: Deformation Transfer via Synchronized Shape Deformation Spaces
• Automatic 3D Mapping for Tree Diameter Measurements in Inventory Operations
• Unsupervised Learning of Intrinsic Structural Representation Points
• PC2WF: 3D Wireframe Reconstruction from Raw Point Clouds
• FlatCAD: Fast Curvature Regularization of Neural SDFs for CAD Models
• Deep Implicit Templates for 3D Shape Representation
• Pose Estimation of Buried Deep-Sea Objects using 3D Vision Deep Learning Models
• Engineering Sketch Generation for Computer-Aided Design
• Dynamical Pose Estimation
• Aligning Instance-Semantic Sparse Representation towards Unsupervised Object Segmentation and Shape Abstraction with Repeatable Primitives
• Theoretical and Empirical Analysis of a Fast Algorithm for Extracting Polygons from Signed Distance Bounds