Neural Random Subspace (NRS), 3D Part

This repo includes the content that we use to verify the effectiveness of NRS (Neural Random Subspace) module on 3D Point Cloud Recognition Task.

It is based on the following implementations:

• https://github.com/WangYueFt/dgcnn
• https://github.com/yanx27/Pointnet_Pointnet2_pytorch

Major implementations of NRS can be found here: https://github.com/CupidJay/NRS_pytorch

For the details of ideas and results, please refer to our paper. We'd love you to cite it if you find it helpful :)

@article{NRS,
	title        = {Neural random subspace},
	author       = {Yun-Hao Cao and Jianxin Wu and Hanchen Wang and Joan Lasenby},
	year         = 2021,
	journal      = {Pattern Recognition},
	volume       = 112,
	pages        = 107801,
	doi          = {https://doi.org/10.1016/j.patcog.2020.107801},
	issn         = {0031-3203}
}

To start with:

- Data Preparation:

bash archive_bash/download_data.sh

- Training Models:

bash archive_bash/train_pointnet.sh
bash archive_bash/train_pointnet2.sh
bash archive_bash/train_dgcnn.sh

- FLOPS and #Params

see utils/FLOPs_Calculator.py for details

- Inference Time

bash archive_bash/timer.sh