git clone https://github.com/Rhoana/topological_thinning.git
cd topological_thinning
conda create -n skeleton_env --file requirements.txt
source activate skeleton_env
cd skeletonization
python setup.py build_ext --inplace
cd ../transforms
python setup.py build_ext --inplace
All datasets are referenced using a meta file. The meta file should have the format meta/{PREFIX}.meta where {PREFIX} is a unique identifier per dataset. The meta file needs to have the following format:
# resolution in nm
6x6x30
# segmentation filename
segmentations/SNEMI3D.h5 main
# grid size
1024x1024x100
where the resolution represents the imaging resolution of the dataset, the segmentation filename is a path to the dataset with the accompanying h5 dataset name, and the grid size is the number of voxels in each dimension. Resolution and Grid Sizes are (x, y, z) format.
There is an example script at examples/generate_skeleton.py.