molierflower/neuralpoints

【Code of CVPR 2022 paper】Neural Points: Point Cloud Representation with Neural Fields for Arbitrary Upsampling

Neural Points

【Code of CVPR 2022 paper】Neural Points: Point Cloud Representation with Neural Fields for Arbitrary Upsampling (CVPR 2022).

• Paper address: https://arxiv.org/abs/2112.04148
• Project webpage: https://wanquanf.github.io/NeuralPoints.html

Prerequisite Installation

The code has been tested on Ubuntu 18, with Python3.8, PyTorch 1.6 and Cuda 10.2:

conda create --name NePs

conda activate NePs

conda install pytorch=1.6.0 torchvision=0.7.0 cudatoolkit=10.2 -c pytorch

conda install -c conda-forge igl

Before running the code, you need to build the cuda&C++ extensions of Pytorch:

cd [ProjectPath]/model/model_for_supp/pointnet2

python setup.py install

How to use the code:

Download our dataset: dataset, (extracting code: qiqq). Put the 'Sketchfab2' folder into: [ProjectPath]/data.

Firstly, you need to change the working directory:

cd [ProjectPath]/model/conpu_v6

To obtain the testing results of the testing set, run:

python train_script101_test.py

To train our network, run:

python train_script101.py

Citation

Please cite this paper with the following bibtex:

@inproceedings{feng2022np,
    author    = {Wanquan Feng and Jin li and Hongrui Cai and Xiaonan Luo and Juyong Zhang},
    title     = {Neural Points: Point Cloud Representation with Neural Fields for Arbitrary Upsampling},
    booktitle = {{IEEE/CVF} Conference on Computer Vision and Pattern Recognition (CVPR)},
    year      = {2022}
}

Acknowledgement

In this repo, we borrowed the backbone structure from DGCNN.