Skip to content

wanglikun1973/SRCNN-keras

 
 

Repository files navigation

Keras implementation of SRCNN

The original paper is Learning a Deep Convolutional Network for Image Super-Resolution

My implementation have some difference with the original paper, include:

  • use Adam alghorithm for optimization, with learning rate 0.0003 for all layers.
  • Use the opencv library to produce the training data and test data, not the matlab library. This difference may caused some deteriorate on the final results.
  • I did not set different learning rate in different layer, but I found this network still work.
  • The color space of YCrCb in Matlab and OpenCV also have some difference. So if you want to compare your results with some academic paper, you may want to use the code written with matlab.

Use:

Create your own data

open prepare_data.py and change the data path to your data

Excute: python prepare_data.py

training and test:

Excute: python main.py

Result(training for 200 epoches on 91 images, with upscaling factor 2):

Results on Set5 dataset:

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%