Neural Network Denoising with the Noise2Void Algorithm

This repository provides a step-by-step guide for training and testing a U-Net Neural Network for image denoising using:

• The JUMP Dataset.
• The BSD68 Benchmark Dataset

The network implements the Noise2Void (N2V) algorithm, as available through the CAREamics library. This approach allows for denoising without requiring ground truth clean images, making it particularly suited for applications where clean references are unavailable.

Project Goals

• Explore the potential of Deep Neural Networks (DNNs) for image denoising.
• Compare results of N2V with those obtained using traditional denoising methods.
• Understand the inner workings of the N2V algorithm and its implementation details.

Steps to Reproduce

1. Setup a Virtual Environment using Anaconda

• Make sure Python and Anaconda ("conda") are installed and working.

• Setup the Conda Environment:

  conda env create -f env/conda.yml

  conda activate n2v

  pip install -r env/requirements.txt

2. Run the scripts

• Follow the Jupyter Notebooks to Train the Model and Generate some Predictions:

  jump_cells.ipynb

  bsd68.ipynb