Analyzing the Sample Complexity of Self-Supervised Image Reconstruction

Each folder contains the code to reproduce the results in one of the Figures 1,2,3,5,6 in the main body of the paper Analyzing the Sample Complexity of Self-Supervised Image Reconstruction Methods.

In particular

• Figure 1: Simulations for subspace denoising
• Figure 2: Gaussian image denoising
• Figure 3: Real-world camera noise denoising
• Figure 5: Compressive sensing for natural images
• Figure 6: Compressive sensing accelerated MRI

Requirements

CUDA-enabled GPU is necessary to run the code. We tested this code using:

• Ubuntu 20.04
• CUDA 11.5
• Python 3.7.11
• PyTorch 1.10.0

Installation

First, install PyTorch 1.10.0 with CUDA support following the instructions here. Then, to install the necessary packages run

pip install -r requirements.txt

We used the bart toolbox to pre-compute the sensitivity maps for the experiments on accelerated MRI. Install bart toolbox by following the instructions on their home page.

Datasets

ImageNet

ImageNet is an open dataset, and you can request access at https://image-net.org/download.php. To run the experiments from our paper, you need to download the ImageNet train set.

Smartphone Image Denoising Dataset (SIDD)

SIDD is an open dataset, and can be donwloaded from https://www.eecs.yorku.ca/~kamel/sidd/. To run the experiments from our paper you need to download the SIDD-Medium dataset, the SIDD validation data and the SIDD benchmark data as it contains the meta data for the validation set.

fastMRI

FastMRI is an open dataset, however you need to apply for access at https://fastmri.med.nyu.edu/. To run the experiments from our paper, you need to download the fastMRI brain dataset.

Acknowledgments and references

The code for MRI reconstruction partly builds on the fastMRI repository, and the code for image denoising on Robust And Interpretable Blind Image Denoising Via Bias-Free Convolutional Neural Networks.

• Klug et al. "Analyzing the Sample Complexity of Self-Supervised Image Reconstruction Methods". In https://arxiv.org/abs/2305.19079 (2023).
• Zbontar et al. "fastMRI: An Open Dataset and Benchmarks for Accelerated MRI". In: https://arxiv.org/abs/1811.08839* (2018).
• Russakovsky et al. "ImageNet Large Scale Visual Recognition Challenge". In: International Journal of Computer Vision (2015).
• Abdelhamed et al. "A High-Quality Denoising Dataset for Smartphone Cameras". In IEEE Computer Vision and Pattern Recognition (CVPR) (2018).