Deep-Zooming-of-Images using SRGAN
Description: The folder contain two Python scripts
Zoom.py : Takes jpg images from Input folder,slice the input image into hundred pieces and convert each sliced images into high resolution images using SRGAN pretrained models. Library used : OpenCV, tensorflow Python Libs included : glob, numpy, image_slicer, os, math, numpy
join_images.py : Join each high resolution slices and reconstruct the zoomed form of orginal image. Library used : OpenCV Python Libs included : numpy
The folder contain one .sh command language interprete
- Zoom.sh : Contains parameters given to Zoom.py
Developed Based On :
- Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network- https://arxiv.org/pdf/1609.04802.pdf
- The code is highly inspired by : pix2pix-tensorflow .
- Pretrained model : https://drive.google.com/uc?id=0BxRIhBA0x8lHNDJFVjJEQnZtcmc&export=download
- python2.7 or python 3.6
- tensorflow r1.10 or above version
- OpenCV 4.0.1 or above
How To Run Code:
- Download the zip file Deep-Zooming-of-Images using SRGAN.zip and extract the files in it.
- Download the pretrained model from the link given above.
- After extracting the folder copy the input image to Input Folder(in jpg format only).
- Run the Zoom.sh.
- The Output image will be generated in Zoom folder.
Warning: Do not forget to remove the temporary folder (temp) generated, if the programe stops in between execution. If it is'nt removed it may affect next execution.