Skip to content

Deep Learning based Image Super Resolution using DCGANs in Keras

Notifications You must be signed in to change notification settings

anjishnu/surreal

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Surreal

Surreal is a Deep Learning based super resolution algorithm

The goal for this project is to provide a generic super resolution algorithm for upscaling images. The first implementation is going to use a conditional DCGAN (Deep Convolutional Generative Adversarial Network) so that the generative model operates on perceptual loss - not some kind of hacky euclidean distance metric.

I am reading the papers below but simplifying the model fairly drastically.

Todos

  • Adapt super-resolution code for an RGB dataset
  • Easy to use data-generator API which is dataset independent.

Done

  • Modularize and make the model definition code more object-oriented to make future modifications easier.
  • Get DCGAN working on MNIST
  • Get super-resolution working for MNIST

References

  • Radford, A., Metz, L. and Chintala, S., 2015. Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434.
  • Ledig, C., Theis, L., Huszár, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z. and Shi, W., 2016. Photo-realistic single image super-resolution using a generative adversarial network. arXiv preprint arXiv:1609.04802.

About

Deep Learning based Image Super Resolution using DCGANs in Keras

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages