a simple code for SRCNN using tensorflow

I refer to the code in the paper [Image Super-Resolution Using Deep Convolutional Networks] (https://arxiv.org/abs/1501.00092v3)

model

Requirements

• python 3
• tensorflow-gpu(or tensorflow, but you might need to edit the code)
• numpy
• PIL
• h5py
• matplotlib

Train

use "python solver.py" to train the model, if you want to indicate the data you want to use, just put your data into the directory and indicate your training data path, just like:

python solver.py --data_dir my_training_data --h5_dir my_train_h5_data

more usage can be found in utils/args.py

Test

python solver.py --mode test and just as above mentioned, you can indicate your test data

Others

Thanks for the projects below, and these projects help me a lot:

• https://github.com/tegg89/SRCNN-Tensorflow
• http://mmlab.ie.cuhk.edu.hk/projects/SRCNN.html