PytorchConnectomics / AxonEM-challenge

utility code for AxonEM challenge

Python Library for ERL Evaluation

ERL Evaluation

• Installation

# create a new environment
conda create -n erl-eval python==3.9.0
source activate erl-eval
conda install -c conda-forge -c ostrokach-forge -c pkgw-forge graph-tool

# clone the repo
git clone --recursive https://github.com/PytorchConnectomics/AxonEM-challenge.git

# install [funlib.evaluate](https://github.com/donglaiw/funlib.evaluate)
cd challenge_eval/funlib.evaluate
pip install -r requirements.txt
python setup.py install
cd ..

(Under challenge_eval/ folder)

• AxonEM evaluation: python test_axonEM.py -s seg_axonM.h5 -g axonM_gt_16nm_skel_stats.p -c 5

Generate Skeleton

• install kimimaro

pip3 install kimimaro 

(Under challenge_eval/ folder)

• GT skeleton generation: python skeleton.py -s snemi_train-labels.tif -r 30x6x6 -i 1,2,3 -o snemi_skel.p
• ERL evaluation: python test_volume.py -s pred_seg.tif -g snemi_skel.p -gu physical -gr 30x6x6