PolyGNN: Polyhedron-based Graph Neural Network for 3D Building Reconstruction from Point Clouds

TL;DR

PolyGNN is a novel graph neural network that reconstructs 3D building models from point clouds by assembling polyhedral primitives, enabling efficient and accurate large-scale city reconstructions.

Contribution

The paper introduces PolyGNN, a polyhedron-based GNN with a skeleton sampling strategy for 3D building reconstruction from point clouds, addressing arbitrary shapes and large-scale data.

Findings

Effective reconstruction of 3D buildings with watertight, compact models.

High transferability across different cities and real-world data.

Demonstrated efficiency and accuracy on large-scale datasets.

Abstract

We present PolyGNN, a polyhedron-based graph neural network for 3D building reconstruction from point clouds. PolyGNN learns to assemble primitives obtained by polyhedral decomposition via graph node classification, achieving a watertight and compact reconstruction. To effectively represent arbitrary-shaped polyhedra in the neural network, we propose a skeleton-based sampling strategy to generate polyhedron-wise queries. These queries are then incorporated with inter-polyhedron adjacency to enhance the classification. PolyGNN is end-to-end optimizable and is designed to accommodate variable-size input points, polyhedra, and queries with an index-driven batching technique. To address the abstraction gap between existing city-building models and the underlying instances, and provide a fair evaluation of the proposed method, we develop our method on a large-scale synthetic dataset with well-defined ground truths of polyhedral labels. We further conduct a transferability analysis across cities and on real-world point clouds. Both qualitative and quantitative results demonstrate the effectiveness of our method, particularly its efficiency for large-scale reconstructions. The source code and data are available at https://github.com/chenzhaiyu/polygnn.