Super-Resolution-Image-Project

This is a web-app trained on SRCNN model. On giving an image as input, it reconstructs a higher resolution image of the same.

Note: Only supports .bmp images

Installation

To run on locally:

• Clone the repository
• Open the application.py file, in the folder.
• Change the

         UPLOAD_FOLDER = '/app/static/input/'
         OUTPUT_FOLDER = '/app/static/output/'
  

  to the location of your static/input and static/output folder respectively.

• Save the file.
• Open the terminal, and create a virtual environment inside the application.
• Run the following commands :

       pip3 install -r requirements.txt
  

  To install all the modules and dependencies.

• To run the application:

         python3 application.py
  

If everything works, the screen will display as follows

Browse and use images, their are few .bmp images added in the source folder of the repository, so those can also be used to test

Working output:

To run the docker container:

• Clone the repository
• Open the terminal, and open the directory of the repository
• Run the following command :

         sudo docker build --tag super-resolution-image-app .
  

  To build the docker image

• To run the docker container in daemon mode with ports exposed :

      sudo docker run --name super-resolution-image-app -p 5001:5001 super-resolution-image-app
  

If everything works, the screen will display as follows



Browse and use images, their are few .bmp images added in the source folder of the repository, so those can also be used to test

Working output:



The model is based on the following paper: https://arxiv.org/abs/1501.00092