Prerequisites:

1. Anaconda 2 or Miniconda 2 (https://conda.io/miniconda.html)
2. CUDA 8 and cudNN 5.1 (https://www.tensorflow.org/versions/r1.2/install/install_linux)

Build a virtual environment

conda env create -f environment.yml

Usage

<!-- Activate the environment -->
source activate dlbi
<!-- Run the code -->
python main.py --mode test --input path/to/lr/images --output output/folder
<!-- Deactivate the environment -->
source deactivate

Explanation of path

There are two modes: test and test_full. In the test mode, we only need to construct one image from 200 lr images. All the 200 images should be put into the same folder. In the test_full mode, we would perform a lot of such tasks. Each 200 images would have one separate folder, which is under the folder we give the program as the input argument.

Input images

Each low resolution image should be one channel, 60 by 60 grayscale image.