implement the paper 'Deep Residual Network for Joint Demosaicing and Super-Resolution' with Keras
use prepare.py
to split images to 64x64x1 and 128x128x3 to get the X and Y
with the train set, we can use train.py
to train our model
For validating,we just need to use prepare_test.py
resize images and get the bayer value.
Use validate.py
to predict image and caculate CSPNR.