Point2CAD: Reverse Engineering CAD Models from 3D Point Clouds

TL;DR

Point2CAD introduces a hybrid analytic-neural approach for reconstructing structured CAD models from 3D point clouds, significantly improving topology accuracy and setting new benchmarks in CAD reconstruction.

Contribution

It presents a novel implicit neural surface representation and a hybrid reconstruction scheme that enhances CAD model recovery from point clouds, outperforming existing methods.

Findings

Achieves state-of-the-art results on the ABC CAD benchmark.

Demonstrates improved topology and surface accuracy.

Compatible with various segmentation backbones.

Abstract

Computer-Aided Design (CAD) model reconstruction from point clouds is an important problem at the intersection of computer vision, graphics, and machine learning; it saves the designer significant time when iterating on in-the-wild objects. Recent advancements in this direction achieve relatively reliable semantic segmentation but still struggle to produce an adequate topology of the CAD model. In this work, we analyze the current state of the art for that ill-posed task and identify shortcomings of existing methods. We propose a hybrid analytic-neural reconstruction scheme that bridges the gap between segmented point clouds and structured CAD models and can be readily combined with different segmentation backbones. Moreover, to power the surface fitting stage, we propose a novel implicit neural representation of freeform surfaces, driving up the performance of our overall CAD reconstruction scheme. We extensively evaluate our method on the popular ABC benchmark of CAD models and set a new state-of-the-art for that dataset. Project page: https://www.obukhov.ai/point2cad}{https://www.obukhov.ai/point2cad.