You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For training a SRGAN HR (high res) and LR (low res) images are needed.
In your code you write training images and validation images.
Do training images correspond to LR and validation images correspond to HR?
Then what do test images correspond to?
The text was updated successfully, but these errors were encountered:
I think I understood this.
Training images are the images you train on. obviously.
This dataset consists of HR respectivly gaussian downsampled images (LR).
Validation images are images of the same class/type as the training images. But those images were not shown to the GAN. Those images are not shown to the GAN during training, so that the GAN is able to generalize better. Validation images consist just like Training images of HR and downsampled LR images.
Test images are images with a native resolution that corresponds to the LR resolution of the trained images and an accuracy is calculated, measuring the quality of the metrics.
For training a SRGAN HR (high res) and LR (low res) images are needed.
In your code you write
training images
andvalidation images
.Do training images correspond to LR and validation images correspond to HR?
Then what do test images correspond to?
The text was updated successfully, but these errors were encountered: