The Devil is in the Pose: Ambiguity-free 3D Rotation-invariant Learning via Pose-aware Convolution
by Ronghan Chen, Yang Cong
in CVPR, 2022. Arxiv

If you have any question about the code or the paper, don't hesitate and open an issuse!😉

Python=3.7, CUDA=10.0, PyTorch=1.4.0, torch_geometric=1.6.0, torch_cluster=1.5.4, torch_sparse=0.6.1, torch_scatter=2.0.4, tensorboardX, scikit-learn, numpy, termcolor

First, please download the ModelNet40 dataset from here(1.6G), and place it at dataset/modelnet40_normal_resampled.

To train a model under SO(3) or z rotations:

sh scripts/PaRINet_so3.sh 
sh scripts/PaRINet_rot_z.sh 

Best model will be saved at log/PaRINet_***/best. And you can evaluate them by running:

sh scripts/test_PaRINet_rot_z.sh
sh scripts/test_PaRINet_so3.sh

To visualize the training process, please run:

tensorboard --logdir log

If you find this work useful, please citing the paper:

@inproceedings{chen2022devil,
  title={The Devil is in the Pose: Ambiguity-free 3D Rotation-invariant Learning via Pose-aware Convolution},
  author={Chen, Ronghan and Cong, Yang},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7472--7481},
  year={2022}
}

• The code framework is borrowed from 3D-CODED
• The code for classification architecture is borrowed from DGCNN

Code on ScanObjectNN and ShapeNetPart.