On automatic data augmentation for 3D point cloud classification (BMVC 2021)

This repository contains the code for our BMVC 2021 paper: On automatic data augmentation for 3D point cloud classification. You can find the PDF of our paper here.

Dependencies

Following is the suggested way to install these dependencies:

# Create a new conda environment
conda create -n AdaPC python=3.7
conda activate AdaPC
conda install pytorch=1.6.0 torchvision cudatoolkit=10.1 -c pytorch -y
pip install -r requirements.txt

Data

Download the ModelNet40 dataset from here and put in into /data folder.

Training

You can train our model using the command below:

python train.py --data_dir data/modelnet40_ply_hdf5_2048 --gpu 0 --log_dir modelnet40_AdaPC_STRy

The default options have scaling, translation and rotation augmentations.

Acknowledgement

Part of our code is borrowed from PointAugment(Li et al) and DARTS (Liu et al).

Citation

If you find our work useful in your research, please consider citing:

@article{zhang2021on,
  title={On Automatic Data Augmentation for 3D Point Cloud Classification.},
  author={Zhang, Wanyue and Xu, Xun and Liu, Fayao and Zhang, Le and Foo, Chuan-Sheng},
  booktitle={BMVC},
  year={2021}
}