search

sefibk / KernelGAN

Other
337 stars 77 forks source link

Is your training data set the same as your test set #78

Open Lincoln20030413 opened 1 year ago

Lincoln20030413 commented 1 year ago

Thankss for your reply for my last question. I also have another question. Is your training data set the same as your test set because in your code there is only one input path. Thanks a lot.

sefibk commented 1 year ago

It is a unique type of training - if you want to fully understand, please read the paper. In short, the model is trained only on the test image (single image, not a test set). There is no training set. It is valid since the test image has no label, the model learns the patches of the image and realizes how to upscale it.

Lincoln20030413 commented 1 year ago

Thanks a lot. I got it.