ShapeFormer: A Transformer for Point Cloud Completion

Mukund Varma T¹, Kushan Raj¹, Dimple A Shajahan^1,2, M. Ramanathan^2
1Indian Institute of Technology Madras, ²TKM College of Engineering

Project Page | Paper | Completion3D-C

To be released

---

Cite this work

Datasets

We use the MVP and Compeletion3D datasets in our experiments, which are available below:

• MVP
• Completion3D

The pretrained models on Completion3D and MVP dataset are available as follows:

• ShapeFormer_pre-trained

Install Python Denpendencies

cd ShapeFormer
pip install -r requirements.txt

Build PyTorch Extensions

NOTE: PyTorch >= 1.4 of cuda version are required.

cd pointnet2_ops_lib
python setup.py install

cd ..

cd Chamfer3D
python setup.py install

cd ..

cd emd
python setup.py install

Acknowledgements

Some of the code of this repo is borrowed from ChamferDistancePytorch. We thank the authors for their great job!