GANs take a long time to train. I have to train and use a machine learning algorithm (not necessarily a deep learning algorithm) that is able to deblur any image (after training it in about 10 hours: that's why I can't use the GAN network to achieve my goal). The quality of the deblur may be worse than with GAN (in other words, I do not have high expectations on the results of the deblurring, which may be imperfect).
As I use TensorFlow and Keras, any illustration (or implementation) using one of these two libraries would be welcome.Research papers that I could read to understand the algorithm would also be welcome.
I searched Google for "CNN deblurring image" but I didn't find any relevant results. Do you know of any networks that meet my criteria?