https://github.com/zaccharieramzi/tf-didn

An unofficial implementation of the Deep iterative down-up CNN for image denoising
https://github.com/zaccharieramzi/tf-didn

denoising neural-network tensorflow

Last synced: 2 months ago
JSON representation

An unofficial implementation of the Deep iterative down-up CNN for image denoising

• Host: GitHub
• URL: https://github.com/zaccharieramzi/tf-didn
• Owner: zaccharieramzi
• License: mit
• Created: 2020-06-10T16:03:32.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2020-06-12T12:55:16.000Z (about 6 years ago)
• Last Synced: 2025-06-12T13:11:29.669Z (about 1 year ago)
• Topics: denoising, neural-network, tensorflow
• Language: Python
• Homepage: http://openaccess.thecvf.com/content_CVPRW_2019/html/NTIRE/Yu_Deep_Iterative_Down-Up_CNN_for_Image_Denoising_CVPRW_2019_paper.html
• Size: 9.77 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# TensorFlow implementation of the Deep iterative down-up CNN

[![Build Status](https://travis-ci.com/zaccharieramzi/tf-didn.svg?branch=master)](https://travis-ci.com/zaccharieramzi/tf-didn)

The Deep iterative down-up CNN (DIDN) is a network introduced by Songhyun Yu et

al. in "Deep Iterative Down-Up CNN for Image Denoising" CVPR 2019.

If you use this network, please cite their work appropriately.

The official implementation is available [here](https://github.com/SonghyunYu/DIDN)

in Pytorch.

The goal of this implementation in TensorFlow is to be easy to read and to adapt:

- all the code is in one file

- defaults are those from the paper

- there is no other imports than from TensorFlow

Some implementation details were taken from the code and not the paper itself:

- no bias is used in the convolutions

- the number of down-up blocks is set to 6

- the activation of the last convolutional layer of the network is linear