virtualenv my_env --python=python3
source my_env/bin/activate
git clone [email protected]:riveSunder/unet_demo.git
cd unet_demo
pip install -r requirements.txt
Download the (labeled) electron microscopy mitochondria segmentation dataset from https://www.epfl.ch/labs/cvlab/data/data-em/ into folders train
, train_labels
, test
, test_labels
in the data folder. Then use the notebook parse_data
to convert and save the tif image stacks into numpy arrays.
Modify the code as you want, then run training:
python unet_src/unet.py