/SDF-Diffusion

Diffusion-Based Signed Distance Fields for 3D Shape Generation (CVPR 2023)

Primary LanguagePython

SDF-Diffusion

Diffusion-Based Signed Distance Fields for 3D Shape Generation (CVPR 2023)

Paper | Project Page

Requirements

  • pytorch
  • pytorch3d
  • h5py
  • einops
  • scipy
  • scikit-image
  • tqdm
  • point-cloud-utils

Dataset

The preprocessed dataset can be downloaded in Huggingface

The dataset (~13GB for resolution 32, ~50GB for 64) should be unzipped and located like this:

SDF-Diffusion
├── config
├── src
├── ...
├── main.py
data
├── sdf-diffusion
│   ├── sdf.res32.level0.0500.PC15000.pad0.20.hdf5
│   ├── sdf.res64.level0.0313.PC15000.pad0.20.hdf5

To use the dataset, please cite ShapeNet:

@article{chang2015shapenet,
  title={Shapenet: An information-rich 3d model repository},
  author={Chang, Angel X and Funkhouser, Thomas and Guibas, Leonidas and Hanrahan, Pat and Huang, Qixing and Li, Zimo and Savarese, Silvio and Savva, Manolis and Song, Shuran and Su, Hao and others},
  journal={arXiv preprint arXiv:1512.03012},
  year={2015}
}

The dataset can be used only for non-commercial research and educational purpose.

Training

Comming Soon

Inference & Evaluation

Comming Soon

Citation

@inproceedings{shim2023diffusion,
  title={Diffusion-Based Signed Distance Fields for 3D Shape Generation},
  author={Shim, Jaehyeok and Kang, Changwoo and Joo, Kyungdon},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={20887--20897},
  year={2023}
}