Skip to content

Latest commit

 

History

History
38 lines (25 loc) · 1.7 KB

README.md

File metadata and controls

38 lines (25 loc) · 1.7 KB

Neural Poisson Surface Reconstruction (nPSR)

Paper | Project Page

This repository contains the implementation of the paper:

Neural Poisson Surface Reconstruction (nPSR): Resolution-Agnostic Shape Reconstruction from Point Clouds

We are currently working on a cleaned-up version of this code that includes more documentation and pre-trained weights. If you find our code or paper useful, please consider citing

@article{andrade2023npsr,
  title={Neural Poisson Surface Reconstruction: Resolution-Agnostic Shape Reconstruction from Point Clouds},
  author={Hector Andrade-Loarca, Julius Hege, Daniel Cremers, Gitta Kutyniok},
  journal={arXiv:2308.01766},
  year={2023}
}

Installation

You need to first install all the dependencies. For that you can use anaconda.

You can create an anaconda environment called npsr using

conda env create -f environment.yaml
conda activate npsr

Training - Quick Start

First, download the data from: ShapeNetsem dataset and place the binvox files inside ./data. Then run the script scripts/generate_training_data_shapenet.py to generate the training data, modifying the corresponding paths. After the training data has been generated one can simply train the model by running scritps/train_shapenet.py.

Inference

Use the notebook examples/Inference.ipynb to evaluate the model on different shapes. You can download our pretrained weights from this link.