soziapath / prnet

prnet

Partial Registration Network

Prerequisites

PyTorch=1.0.1: https://pytorch.org (PyTorch 1.1 has a svd bug, which will crash the training)

scipy>=1.2

numpy

h5py

tqdm

sklearn

Conda environment

conda env create -f environment.yml --name prnet

conda activate prnet

Training

exp1 modelnet40

python main.py --exp_name=exp1

exp2 modelnet40 unseen

python main.py --exp_name=exp2 --unseen=True

exp3 modelnet40 gaussian noise

python main.py --exp_name=exp3 --gaussian_noise=True

Citation

Please cite this paper if you want to use it in your work,

@InProceedings{Wang_2019_NeurIPS,
  title={PRNet: Self-Supervised Learning for Partial-to-Partial Registration},
  author={Wang, Yue and Solomon, Justin M.},
  booktitle = {33rd Conference on Neural Information Processing Systems (To appear)},
  year={2019}
}

Code Reference

Code reference: Deep Closest Point

License

MIT License