Skip to content

mlvlab/PointWOLF

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PointWOLF: Point Cloud Augmentation with Weighted Local Transformations

This repository is the implementation of PointWOLF.

Sihyeon Kim1*, Sanghyeok Lee1*, Dasol Hwang1, Jaewon Lee1, Seong Jae Hwang2, Hyunwoo J. Kim1†, Point Cloud Augmentation with Weighted Local Transformations (ICCV 2021).
1Korea University 2University of Pittsburgh

PointWOLF_main

Installation

Dependencies

  • CUDA 10.2
  • Python 3.7.1
  • torch 1.7.0
  • packages : sklearn, numpy, h5py, glob

Download

Clone repository

$ git clone https://github.com/mlvlab/PointWOLF.git

Download ModelNet40

Notes : When you run the main.py, ModelNet40 is automatically downloaded at .../PointWOLF/data/.
If you want to download dataset on your ${PATH}, see below.

$ cd ${PATH}
$ wget https://shapenet.cs.stanford.edu/media/modelnet40_ply_hdf5_2048.zip --no-check-certificate
$ unzip modelnet40_ply_hdf5_2048.zip
$ rm modelnet40_ply_hdf5_2048.zip

Runnig the code

train

  • Run the training without PointWOLF & AugTune:
$ python main.py --exp_name=origin --model=dgcnn --num_points=1024 --k=20 --use_sgd=True
  • Run the training with PointWOLF:
$ python main.py --exp_name=PointWOLF --model=dgcnn --num_points=1024 --k=20 --use_sgd=True --PointWOLF
  • Run the training with PointWOLF & AugTune:
$ python main.py --exp_name=PointWOLF_AugTune --model=dgcnn --num_points=1024 --k=20 --use_sgd=True --PointWOLF --AugTune

eval

  • Run the evaluation with trained model located at ${PATH}:
$ python main.py --exp_name=eval --model=dgcnn --num_points=1024 --k=20 --use_sgd=True --eval=True --model_path=${PATH}

Citation

@InProceedings{Kim_2021_ICCV,
    author    = {Kim, Sihyeon and Lee, Sanghyeok and Hwang, Dasol and Lee, Jaewon and Hwang, Seong Jae and Kim, Hyunwoo J.},
    title     = {Point Cloud Augmentation With Weighted Local Transformations},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2021},
    pages     = {548-557}
}

License

MIT License

Acknowledgement

The structure of this codebase is borrowed from DGCNN.

About

Official Implementation (PyTorch) of "Point Cloud Augmentation with Weighted Local Transformations", ICCV 2021

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages