Creating the dataset

• create two folders A and B with subfolders train and test in each.
• Keep the clean train data in A and the blurred data in B in respective train and test subfolders
• Create the concatenated images using

python scripts/combine_A_and_B.py --fold_ A /path to fold A --fold_B /path to fold B --fold_AB /path to save the catenated image

How to train and test

-Edit the parameters inside train.lua

• For training follow the command

DATA_ROOT=/path_to_catenated_data/  name=expt_name which_direction=BtoA gpu=3 th train.lua

-For testing

 DATA_ROOT=/path/to/data/ name=expt_name which_direction=BtoA phase=val th test.lua

Prerequisites

• Linux or OSX
• NVIDIA GPU + CUDA CuDNN (CPU mode and CUDA without CuDNN may work with minimal modification, but untested)

Requirements

• Install torch and dependencies from https://github.com/torch/distro
• Install torch packages nngraph and display

luarocks install nngraph
luarocks install https://raw.githubusercontent.com/szym/display/master/display-scm-0.rockspec

Acknowledgments

Code borrows heavily from DCGAN. The data loader is modified from DCGAN and Context-Encoder.