This repository contains all files that were used for the blog series 'Creating and training a U-Net model with PyTorch for 2D & 3D semantic segmentation - A guide to semantic segmentation with PyTorch and the U-Net'. You can find the blog posts here.
In requirements.txt you'll find the packages for the conda environment that I used. This does not necessarily mean that these will work on your machine/computer (for example torch comes with a specific cuda version).
I have updated the repo, e.g. the transformations and added a dataset example for a 3D dataset. There is also an example now that shows how you can use a segmentation model like the UNet (or any other segmentation model) in PyTorch Lightning in combination with a logger like neptune.ai for experiment tracking.