This is PyTorch implementation of PU-Net based on Official TF repo punet_tf. The code has been tested with Python 3.6.9, PyTorch 1.2.0 and CUDA 10.2 on a Tesla T4.
We have provided official patched data in HDF5 format for trainning, which are stored in './datas/'. To output upsampled points, 20 objects with 5k points are put into ./datas/test_data/our_collected_data/MC_5k
. For evaluation, test mesh files are downloaded into ./datas/test_data/test_mesh
.
# KNN-CUDA offline installation
pip install KNN_CUDA-0.2-py3-none-any.whl
# Pointnet2
cd extensions/pointnet2
python setup.py install
You can switch 'mode' to train or test. The trained model will be stored in '.weights/punet' and the upsampled points for test will be stored in './results/punet'.
python main.py --mode train
python test.py --mode test
Firstly, you should install CGAL library by running
apt-get update
apt-get upgrade
apt-get install libcgal-dev libcgal-demo
compile cpp code
cd evaluation_code
cmake .
make
generate density file.
cd /evaluation_code
sh evaluate_all.sh
calculate NUC metric
python nuc_utils/calculate_nuc.py
NUC with different p
0.002 | 0.004 | 0.006 | 0.008 | 0.010 | 0.012 | 0.015 |
---|---|---|---|---|---|---|
0.17464873 | 0.14155353 | 0.12641019 | 0.11760626 | 0.11187568 | 0.11156724 | 0.11221573 |
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》