Basic implementation of the U-Net Convolutional Neural Network Architecture, implemented using Keras with a Tensorflow backend
The dependencies list is not exhaustive - I have only mentioned the packages that came to mind. Others are probably needed, but the versioning is likely less important. I also strongly recommend the use of a virtual environment when running this to avoid conflicts, as some of the packages are out-of-date compared to their most recent versions. Make sure to also check the CUDA and CUDNN requirements for Tensorflow if using tensorflow-gpu.
Main dependencies/versions used (for building and running the U-Net):
numpy >=1.14.4
scipy >=1.1.0
tensorflow ==1.2.0 (should work with later versions, but untested)
tensorflow-gpu ==1.2.0 (can ignore if not processing on GPU, but will be slower)
Keras ==2.0.6 (should work with later versions, but untested)
h5py >=2.8.0
Pillow ==5.1.0
Other dependencies:
SimpleITK =0.9.1 (for preprocessing images only)
matplotlib (only for debug purposes, for the purpose of viewing intermediate images)
This project is not actively maintained as I am currently working on other projects, but will likely be extended in the future by either me or other researchers.