Perturbed Self-Distillation: Weakly Supervised Large-Scale Point Cloud Semantic Segmentation (ICCV2021)

This is the implementation of PSD (ICCV 2021), a simple weakly-supervised semantic segmentation of large-scale 3D point clouds.

(1) Setup

This code has been tested with Python 3.5, Tensorflow 1.13, CUDA 9.0 and cuDNN 7.4.1 on Ubuntu 16.04.

• Clone the repository

git clone --depth=1 https://github.com/Yachao-Zhang/PSD && cd PSD

• Setup python environment

pip install -r helper_requirements.txt
sh compile_op.sh

(2) Weakly semantic Segmentation on S3DIS

S3DIS dataset can be found here. Download the files named "Stanford3dDataset_v1.2_Aligned_Version.zip". Uncompress the folder and move it to /data/S3DIS.

• Preparing the dataset:

python utils/data_prepare_s3dis.py

Training and test of weakly semantic Segmentation on S3DIS Area-5 by:

sh jobs_s3dis_a5.sh 

Citation

If you find our work useful in your research, please consider citing:

@inproceedings{zhang2021perturbed,
    title={Perturbed Self-Distillation: Weakly Supervised Large-Scale Point Cloud Semantic Segmentation},
    author={Zhang, Yachao and Qu, Yanyun and Xie, Yuan and Li, Zonghao and Zheng, Shanshan and Li, Cuihua},
    booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
    pages={15520--15528},
    year={2021}
}

A related work (Weakly Supervised Semantic Segmentation for Large-Scale Point Cloud AAAI-2021) can be found here.

@inproceedings{zhang2021weakly,
    title={Weakly Supervised Semantic Segmentation for Large-Scale Point Cloud},
    author={Zhang, Yachao and Li, Zonghao and Xie, Yuan and Qu, Yanyun and Li, Cuihua and Mei, Tao},
    booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
    volume={35},
    number={4},
    pages={3421--3429},
    year={2021}
}

Acknowledgment

Note that this code is heavily borrowed from RandLA-Net (https://github.com/QingyongHu/RandLA-Net).