Modular-U-Net

Customizable U-Net model to be used with any .npy image, with a lot of options to change the internal workings, such as pooling layers used, data augmentation and more settings.

Table of Contents

• Prerequisites
• Installation
• Usage

Prerequisites

For this project you should have Python 3.9+ installed and preferably CUDA to significantly increase the speed of training.

Installation

1. Clone the repository

Clone the respotitory to your desired location.

2. Set up virtual environment

In order to set up manually the virtual environment you need to run the following commands.

1. Run python -m venv unet to create the virtual environment.

2. Nagivate into unet/Scipts and run the activate script.

3. Go to the Pytorch website and find the suitable Pytorch version and install it by running the command provided.

4. Run pip install openpyxl opencv-python to have all required packages.

Usage

1. Activate virtual environment

Nagivate into the scripts folder in the virtual environment and use activate if not already active.

2. Set config.py

Check the config.py file to change any settings required such as data location and result location.

3. Run the model

Use command python model.py to run the model.