Skip to content
/ Super-Resolution_CNN-PyTorch Public

Implementation of 'Image Super-Resolution using Deep Convolutional Network'

License

MIT license
24 stars 4 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

YeongHyeon/Super-Resolution_CNN-PyTorch

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
dataset
dataset
 
 
readme
readme
 
 
runs
runs
 
 
source
source
 
 
LICENSE
LICENSE
 
 
README.md
README.md
 
 
run.py
run.py
 
 

Repository files navigation

[PyTorch] Super-Resolution CNN

PyTorch implementation of 'Image Super-Resolution using Deep Convolutional Network'. TensorFlow version is also provided in Related Repository.

Architecture

The architecture of the Super-Resolution Network (SRCNN).

The architecture constructed by three convolutional layers, and the kernel size are 9x9, 1x1, 3x2 respectively. It used RMS loss and stochastic gradient descent opeimizer for training in this repository, but original one was trained by MSE loss (using same optimizer). The input of the SRCNN is Low-Resolution (Bicubic Interpolated) image that same size of the output image, and the output is High-Resolution.

Results

Reconstructed image in each iteration (1k, 10k, 100k iterations).

Comparison between the input (Bicubic Interpolated), reconstructed image (by SRCNN), and target (High-Resolution) image.

Requirements

  • Python 3.6.8
  • PyTorch 1.2.0
  • Numpy 1.14.0
  • Matplotlib 3.1.1

Reference

[1] Image Super-Resolution Using Deep Convolutional Networks, Chao Dong et al., https://ieeexplore.ieee.org/abstract/document/7115171/
[2] Urban 100 dataset, Huang et al., https://sites.google.com/site/jbhuang0604/publications/struct_sr

First commit: 07.October.2018

About

Implementation of 'Image Super-Resolution using Deep Convolutional Network'

Topics

interpolation architecture pytorch super-resolution deep-convolutional-networks srcnn

Resources

Readme

License

MIT license
Activity

Stars

24 stars

Watchers

2 watching

Forks

4 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages