Superpoint Network for Point Cloud Oversegmentation

by Le Hui, Jia Yuan, Mingmei Cheng, Jin Xie, Xiaoya Zhang, and Jian Yang

This is a rough README.md, and we'll work through it. !!!

Project Code

Requirements

• basic environment

  Python 3.6.6
  Pytorch 1.4.0
  CUDA 10.1
  

• compile the "libply_c" library (Please refer to SPG)

  CONDAENV=YOUR_CONDA_ENVIRONMENT_LOCATION
  cd libs/ply_c
  cmake . -DPYTHON_LIBRARY=$CONDAENV/lib/libpython3.6m.so -DPYTHON_INCLUDE_DIR=$CONDAENV/include/python3.6m -DBOOST_INCLUDEDIR=$CONDAENV/include -DEIGEN3_INCLUDE_DIR=$CONDAENV/include/eigen3
  make
  cd ../../
  

• build the ops

  cd libs/pointops && python setup.py install && cd ../../
  
  Note that this may take a long time.
  

Training and Evaluation

• To train and evaluate SPNet, run the following command:

  # Train & Eval
  # Note that you should change the paths in the yaml file.
  
  sh tool/sh_train.sh s3dis 20220121 config/spnet.yaml
  
  sh tool/sh_test.sh s3dis 20220121 config/spnet.yaml 850
  

Citation

If you find the code or trained models useful, please consider citing:

@inproceedings{hui2021spnet,
  title={Superpoint Network for Point Cloud Oversegmentation},
  author={Hui, Le and Yuan, Jia and Cheng, Mingmei and Xie, Jin and Yang, Jian},
  booktitle={ICCV},
  year={2021}
}

Acknowledgement

Our code refers to SPG and PointWeb. Many thanks to SPG for a great work.