This project is no longer maintained, please use the updated version: https://github.com/NJUVISION/PCGCv2 and our latest work: https://github.com/NJUVISION/SparsePCGC
Jianqiang Wang, Hao Zhu, Haojie Liu, Zhan Ma. [arXiv]
(a) (b) An illustrative overview in (a) and detailed diagram in (b) for point cloud geometry compression consisting of a pre-processing for PCG voxelization, scaling & partition, a compression network for compact PCG representation, and metadata signaling, and post-processing for PCG reconstruction and rendering. “Q” stands for “Quantization”, “AE” and “AD” are Arithmetic Encoder and Decoder respectively. “Conv” denotes the convolution layer with the number of the output, channels, and kernel size.
python test.py compress "testdata/8iVFB/longdress_vox10_1300.ply" \
--ckpt_dir="checkpoints/hyper/a6b3/"
python test.py decompress "compressed/longdress_vox10_1300" \
--ckpt_dir="checkpoints/hyper/a6b3/"
Other default options: --scale=1, --cube_size=64, --rho=1.0, --mode='hyper', --modelname='models.model_voxception'
Please refer to demo.ipynb
for each step.
python eval.py --input "testdata/8iVFB/longdress_vox10_1300.ply" \
--rootdir="results/hyper/" \
--cfgdir="results/hyper/8iVFB_vox10.ini" \
--res=1024
Different parameters are required for different dataset, for example:
python eval.py --input "testdata/Sparse/House_without_roof_00057_vox12.ply" \
--rootdir "results/hyper/" \
--cfgdir "results/hyper/House_without_roof_00057_vox12.ini" \
--res=4096
The detailed cfgs and results can be downloaded in https://box.nju.edu.cn/f/b78aeedc0453442aafe5/ And several examples of decoded point clouds can be download in https://box.nju.edu.cn/d/f6a6f8ae61c94cea9248/
sampling points from meshes, here we use pyntcloud (pip install pyntcloud)
cd dataprocess
python mesh2pc.py
The output point clouds can be download in http://yun.nju.edu.cn/d/227493a5bd/
python generate_dataset.py
the output training dataset can be download in http://yun.nju.edu.cn/d/604927e275/
python train_hyper.py --alpha=0.75 \
--prefix='hyper_' --batch_size=8 --init_ckpt_dir='checkpoints/hyper/a0.75b3' --reset_optimizer=1
or
python train_factorized.py --alpha=2 \
--prefix='voxception_' --batch_size=8 --init_ckpt_dir='./checkpoints/factorized/a2b3' --reset_optimizer=1
results.ipynb
These files are provided by Nanjing University Vision Lab. Thanks for the help from SJTU Cooperative Medianet Innovation Center. Please contact us (wangjq@smail.nju.edu.cn) if you have any questions.