[Super Resolution] opt.mode = 'SR_train'

[marisanest]

Hi,

I try to understand, how the super resolution training works. As you already mentioned, it is just the same procedure as for normal training but with a up-sampled image. If I look at the code within training.py, I see multiple code parts like opt.mode == 'SR_train', but it seems that these code parts are never executed due to the fact, that within the SR.py file the opt.mode is set to train before the actual training.

So my question is: Is this code within training.py redundant or am I missing something here?

Thanks!

[tamarott]

opt.mode == 'SR_train' is a flag I used during the experimental phase.
it is not needed for the final version.

Super resolution training is the same as regular training, but the scale factor between different model scales is adjusted according to the desired SR scale.

[marisanest]

Ok great, thank you for your fast response! :)