Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network implemented in Keras

Architecture:

Generator and Discriminator Network:

Perceptual Loss

Adversarial Loss

Content Loss

Network Details:

* 16 Residual blocks used.
* PixelShuffler x2: This is feature map upscaling. 2 sub-pixel CNN are used in Generator.
* PRelu(Parameterized Relu): We are using PRelu in place of Relu or LeakyRelu. It introduces learn-able parameter 
  that makes it possible to adaptively learn the negative part coefficient.
* k3n64s1 this means kernel 3, channels 64 and strides 1.
* Loss Function: We are using Perceptual loss. It comprises of Content(Reconstruction) loss and Adversarial loss.

How to use

python3 test.py

Output

High Resolution

Low Resolution

Prediction

Reference:

Keras-SRGAN
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network