A PyTorch 1.1.0 implementation of SRCNN and EDSR neural networks for Super Resolution Single Frame tasks.
Since the documentation is a WIP I suggest taking a look at the following files to figure out how to use my code:
-
init_data.py
: starting from source data it performs some preprocessing (eg. RGB-to-GreyScale) and organizes examples and labels into a folder hierarchy. -
init_models.py
: performs training and testing of the two models and collects performance logs during all the operations. -
benchmark.py
: using pre trained models and performance data generated by the previous script, it compares the models and plots useful informations.