Official Code for Paper ``Improving Adversarial Robustness of 3D Point Cloud Classification Models''(ECCV 2022)

Requirements

• Tensorflow>=1.14.0 (not support Tensorflow 2.0)
• Pytorch>=1.2.0
• PointCutMix-K

Compiling Cuda Operations

Please follow this repo.

Dataset

The ModelNet40 can be downloaded from here.

Training and Evaluating

Pretrained model can be found in here. Or, you can train your own model from scratch.

python train.py

Acknowledgment

Parts of code are from DGCNN, PointCloud-Saliency-Map and PointCutMix-K.

Cite

If you find our work is useful, please cite it with the following format:

@inproceedings{li2022improving,
  title={Improving Adversarial Robustness of 3D Point Cloud Classification Models},
  author={Li, Guanlin and Xu, Guowen and Qiu, Han and He, Ruan and Li, Jiwei and Zhang, Tianwei},
  booktitle={European Conference on Computer Vision},
  pages={672--689},
  year={2022},
  organization={Springer}
}