BSP-Net: Generating Compact Meshes via Binary Space Partitioning

By Zhiqin Chen, Andrea Tagliasacchi. Computer Vision and Pattern Recognition, 2019

362 citations recorded by the source corpus.

Polygonal meshes are ubiquitous in the digital 3D domain, yet they have only played a minor role in the deep learning revolution. Leading methods for learning generative models of shapes rely on implicit functions, and generate meshes only after expensive iso-surfacing routines. To overcome these challenges, we are inspired by a classical spatial data structure from computer graphics, Binary Space Partitioning (BSP), to facilitate 3D learning. The core ingredient of BSP is an operation for recursive subdivision of space to obtain convex sets. By exploiting this property, we devise BSP-Net, a network that learns to represent a 3D shape via convex decomposition. Importantly, BSP-Net is unsupervised since no convex shape decompositions are needed for training. The network is trained to reconstruct a shape using a set of convexes obtained from a BSP-tree built on a set of planes. The convexes inferred by BSP-Net can be easily extracted to form a polygon mesh, without any need for iso-surfacing. The generated meshes are compact (i.e., low-poly) and well suited to represent sharp geometry; they are guaranteed to be watertight and can be easily parameterized. We also show that the reconstruction quality by BSP-Net is competitive with state-of-the-art methods while using much fewer primitives. Code is available at https://github.com/czq142857/BSP-NET-original.

Subfields: Computational Mechanics, Computer Graphics and Computer-Aided Design

View Paper: https://www.semanticscholar.org/paper/40c97a8139c497f5417fde8c48a3efe1e6035bc2

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Graphics and Computer-Aided Design

Citations in the corpus

DeepCAD: A Deep Generative Network for Computer-Aided Design Models (2021, methodology): “However, existing 3D generative models merely create computer discretization of 3D shapes: 3D point clouds [6, 52, 53, 8, 30], polygon meshes [17, 42, 31], and levelset ﬁelds [12, 33, 29, 50, 11].”
InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models (2024, background): “Early attempts on image-to-3D mainly focus on the single-view reconstruction task [4, 29, 33, 34, 50, 62].”
Canonical Capsules: Unsupervised Capsules in Canonical Pose (2020, methodology): “We perform decomposition at a single level, and it would be interesting to investigate how to effectively engineer multi-level decompositions [51]; one way could be to over-decompose the input in a redundant fashion (with large K), and use a downstream layers that “selects” the decomposition heads t”
Neural Geometric Level of Detail: Real-time Rendering with Implicit 3D Shapes (2021, background): “BSP-Net and CvxNet [6, 10] learn implicit geometry with space-partitioning trees.”
Grasping Field: Learning Implicit Representations for Human Grasps (2020, background): “Learning 3D object models using various types of representations has also been explored [57–64].”
MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (2023, methodology): “Mesh generation has been approached with various learning-based methods [7, 11, 18, 43].”