NeurCADRecon: Neural Representation for Reconstructing CAD Surfaces by Enforcing Zero Gaussian Curvature

Project | Paper

This repository is the official PyTorch implementation of our paper, NeurCADRecon: Neural Representation for Reconstructing CAD Surfaces by Enforcing Zero Gaussian Curvature.

This code is based on the SIREN, we also provide the implementation based on the IGR: NeurCADRecon-IGR

News

🔥 This paper was accepted by ACM TOG (SIGGRAPH 2024)
⭐ July 29, 2024 (GMT -7): Gave a talk at SIGGRAPH 2024 on NeurCADRecon.

Requirements

python 3.7
CUDA 12.2
pytorch 1.13.0

Installation

git clone https://github.com/QiujieDong/NeurCADRecon.git
cd NeurCADRecon

Preprocessing

Sampling and normalizing to [-0.5, 0.5]

cd pre_processing
python pre_data.py

gt_path: The ground truth mesh of the CAD model.
input_path: The input point cloud that need to be reconstructed.

Overfitting

cd surface_reconstruction
python train_surface_reconstruction.py

All parameters are set in the surface_recon_args.py.

Cite

If you find our work useful for your research, please consider citing the following papers :)

@article{Dong2024NeurCADRecon,
author={Dong, Qiujie and Xu, Rui and Wang, Pengfei and Chen, Shuangmin and Xin, Shiqing and Jia, Xiaohong and Wang, Wenping and Tu, Changhe},
title={NeurCADRecon: Neural Representation for Reconstructing CAD Surfaces by Enforcing Zero Gaussian Curvature},
journal={ACM Transactions on Graphics},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
year={2024},
month={July},
volume = {43},
number={4},
doi={10.1145/3658171},
keywords = {CAD model, unoriented point cloud, surface reconstruction, signed distance function, Gaussian curvature}
}

Acknowledgments

Our code is inspired by Neural-Singular-Hessian, SIREN, and IGR.

sysuyl/NeurCADRecon