Learning Invariant Representation for Unsupervised Image Restoration (CVPR 2020)

Introduction

This is an implementation for the paper "Learning Invariant Representation for Unsupervised Image Restoration" (CVPR 2020), a simple and efficient framework for unsupervised image restoration, which is injected into the general domain transfer architecture. More details could be found in the original paper.

Network Architecture

Prerequisites

• (OS) Windows/Ubuntu
• Python >= 3.6
• Pytorch >= 1.1.0
• Python-Libs, e.g., cv2, skimage.

Training

• Prepare your dataset. In our experiments, we used the PascalVoc dataset to generate training data for Gaussian noise removal.
• Generate Gaussian or Poisson noise via skimage-lib.
• Update the data paths in config.py and utils.py file.
• Train your model by the train.py file.

Test

A simple script to test your model:

python3 test.py

Results

• Gaussian Noise Removal

• Poisson Noise Removal

• Medical Image Denoising (Low-Dose CT)

Extending for other IR tasks

You could extend this work for other image restoration tasks, e.g., super-resolution, deblurring and so on. If so, you need to adjust some hyperparameters for them, and extra self-supervised modules also need to be altered. In this paper, we just provide a more general idea to process the unsupervised image restoration tasks via representation learning.

Acknowledge

Our code is based on the UNIT, which is a nice work for unsupervised image translation.