3DVerifier: efficient robustness verification for 3D point cloud models

Supplementary work

In the supplementary_work.pdf, we demonstrate the power of JANet and provide the network configuration.

First build the enviroment:

conda create --name cnncert python=3.6 source activate cnncert conda install pillow numpy scipy pandas h5py tensorflow numba posix_ipc matplotlib The Tensorflow version should below tf1.15

Then download our model checkpoints

https://livelancsac-my.sharepoint.com/:f:/g/personal/mur2_lancaster_ac_uk/Eirostdd_-tOjpZmkU-yFdkBF6auqdp3IgWDur3ZcTnkyg?e=3fUXhe

To obtain the average bounds of 64 points on 12 layes with average pooling in PointNet model with T-Net, you can run

python main.py

To obtain the distortion from attack method, you could run

python atmain.py

Note: This work is accepted by Machine Learning Journal. Pls find the paper here: 3DVerifier: efficient robustness verification for 3D point cloud models

-- Ronghui Mu & Wenjie Ruan