Francesco

If you have an issue with a str has no attribute decode(), try first doing this: `pip install h5py==2.10.0 --force-reinstall` it's being added to the requirements and still needs to...

This is equivalent to the MSE part. It is a badly setup example. It should be 1.0 at the start of training and then should be reduced in favor of...

Hi, session-config.yml is only a summary of the training session you trigger, that is useful to resume the state of training sessions and to keep track of what you';ve done....

Hi, currently we do not plan to provide benchmarks on any hardware. However: - latency is a known issue with image super-resolution in general. The computational cost is very high...

Can you reproduce this? It looks like an issue with the download of the VGG model from Keras. It is in the middle of your training session, so I would...

This is normal if you did not compile Tensorflow yourself. That should not cause any issue in principle. Could you post your code snippet?

I suspect it might be a bias from the dataset which contains a lot of natural images. Would be interesting to see if running something like a color normalization through...

Hi, the first few epochs (~20) you might want to train the model only with the MAE instead than adversarial loss (maybe without adding noise?). There is not automatic switch...

If you have an adequate dataset, I would use a 4x upscaling. For the rest, your configuration looks fine. In my experiment I observed no to very little qualitative difference...

Nice idea, but that kinda means to go back in time. Image restoration used to be the pre-2015(16?) approach, which according to the literature delivers worse results. Nevertheless, give it...