Implementation of some CNN models using Keras: The Python Deep Learning library

Implementation of FSRCNN model based on Accelerating the Super-Resolution Convolutional Neural Network

Implementation of IRCNN model Keras based on Learning Deep CNN Denoiser Prior for Image Restoration

This repository contains some pre-trained models that can be enabled by creating symbolic links to the checkpoint directory and the configuration file in the root path. This way the code will detect these files and will load the weights of the corresponding model.

ln -s saved_checkpoints/FSRCNN/checkpoint
ln -s saved_checkpoints/FSRCNN/config.ini

ln -s saved_checkpoints/IRCNN/checkpoint
ln -s saved_checkpoints/IRCNN/config.ini

python main.py --train --data /path/to/<training images dir> --val /path/to/<validation images dir> -c config.ini

python main.py --test /path/to/<test images dir> --output /path/to/<output dir> -c config.ini