theanonSR

theanonSR

Super Resolution using Deep Convolutional Neural Network (SRCNN) using theano.

See github page for detail.

Introduction

theanonSR upscales photo image to x2 size.

Original image

slide

Upscaled image using python OpenCV library

slide

Upscaled image using theanonSR

slide

 

 

References

Originally, I was inspired this project from waifu2x project, which uses Torch7 to implement SRCNN.

  • Chao Dong, Chen Change Loy, Kaiming He, Xiaoou Tang, “Image Super-Resolution Using Deep Convolutional Networks”, http://arxiv.org/abs/1501.00092

    SRCNN, super resolution using deep convolutional neural network, is introduced in this paper.

  • waifu2x

    It is the popular project for image super resolution for Anime-Style art. It also has a good performance.

  • theano

    Machine learning library which can be written in python. It also provides nice Deep Learning Tutorials to study how to implement deep neural network.

 

 

Contribution is welcome

The performance of SR for this project is not matured. You are welcome to improve & contribute this project. If you could get any model which performs better performance, feel free to send me a pull request!

Sponsored Links

Leave a Reply

Your email address will not be published.