Skip to content

Cupy-based

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

Cupy-based

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
__pycache__
__pycache__
 
 
apps
apps
 
 
assets
assets
 
 
build
build
 
 
data
data
 
 
result
result
 
 
results
results
 
 
src
src
 
 
README.md
README.md
 
 
runAGR.py
runAGR.py
 
 
runply.py
runply.py
 
 
runxyz.py
runxyz.py
 
 
utils_pgr.py
utils_pgr.py
 
 

README.md

AGR

Environment and Dependencies

This program has been tested on Ubuntu 20.04 with Python 3.8 and CUDA 11.1. This program requires these Python packages: numpy, cupy, scipy and tqdm.

Compiling Source Files

This program uses third-party libraries CLI11 and cnpy. We provide copies of them in this repository for convenicence. If you have cmake, you can build with:

mkdir build
cd build
cmake ..
make -j8

Or you can also build with g++ directly:

cd src
g++ PGRExportQuery.cpp Cube.cpp Geometry.cpp MarchingCubes.cpp Mesh.cpp Octnode.cpp Octree.cpp ply.cpp plyfile.cpp cnpy/cnpy.cpp -ICLI11 -o ../apps/PGRExportQuery -lz -O2
g++ PGRLoadQuery.cpp Cube.cpp Geometry.cpp MarchingCubes.cpp Mesh.cpp Octnode.cpp Octree.cpp ply.cpp plyfile.cpp cnpy/cnpy.cpp -ICLI11 -o ../apps/PGRLoadQuery -lz -O2

If successful, this will generate two executables PGRExportQuery and PGRLoadQuery in ParametricGaussRecon/apps. The former builds an octree and exports grid corner points for query; the latter loads query values solved by PGR and performs iso-surfacing.

(For ease of users, we have compiled and can directly use the two files mentioned above (Linux system). Please grant them executable permissions.)

All code must run on a GPU, only the GPU version is provided here.(It can be run directly on a Linux server (ensuring that PGRExportQuery and PGRLoadQuery have set read-write permissions))

python runAGR.py data/xyz/bunny_10000.xyz --alpha 2 -wk 16	

This will create a folder `results/bunny_10000.xyz` together with three subfolders: `recon`, `samples` and `solve` :

- `recon` contains the reconstructed meshes in PLY format.
- `samples` contains the input point cloud in XYZ format, the normalized input point cloud and the octree grid corners as query set in NPY format.
- `solve` contains the solved Linearized Surface Elements in XYZ and NPY formats, the queried values, query set widths in NPY format, and the iso-value in TXT format.
python runAGR.py data/xyz/Utah_teapot_3000.xyz --alpha 2 -wk 16

For the convenience of operation, we provide a version that can be directly reconstructed from. py (randomly selected points from the mesh), with a default selection of 5K points in this version.

python runply.py data/ply/bunny.ply --alpha 2 -wk 16 python runply.py data/ply/light.ply

More files(.ply) can be obtained by run:

python download_datasets_abc.py python download_datasets_thingi10k.py python download_datasets_real_world.py python download_datasets_famous.py in /data