This repository contains the implementation of Convolutional Occupancy Networks for Point Clouds with Visibility Information as described in the paper Deep Surface Reconstruction for Point Clouds with Visibility Information.

The code is largely based on the original repository.

Data

The datasets used in this repository can be downloaded here.

The pretrained models can be downloaded with:

bash scripts/download_pretrained.sh

Reconstruction

For reconstructing e.g. the ModelNet10 dataset run

python generate.py configs/pointcloud/modelnet/config

where config should be replaced with

• modelnetTR.yaml for reconstruction from a point cloud (traditional ConvONet)
• modelnetSV.yaml for reconstruction from a point cloud augmented with sensor vectors
• modelnetAP.yaml for reconstruction from a point cloud augmented with sensor vectors and auxiliary points

References

If you find the code or data in this repository useful, please consider citing

@misc{sulzer2022deep,
      title={Deep Surface Reconstruction from Point Clouds with Visibility Information}, 
      author={Raphael Sulzer and Loic Landrieu and Alexandre Boulch and Renaud Marlet and Bruno Vallet},
      year={2022},
      eprint={2202.01810},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

@inproceedings{Peng2020ECCV,
 author =  {Songyou Peng, Michael Niemeyer, Lars Mescheder, Marc Pollefeys, Andreas Geiger},
 title = {Convolutional Occupancy Networks},
 booktitle = {European Conference on Computer Vision (ECCV)},
 year = {2020}}