I’ve heard that images which are used for transfer learning of Resnet (i.e., pre-trained) models should be resized to 224x224px (or other kinds of transformations). I wonder if it is required when using fine_tune
albeit a description in the docs which follows:
If
normalize
andpretrained
areTrue
, this function adds aNormalization
transform to thedls
(if there is not already one) using the statistics of the pretrained model. That way, you won’t ever forget to normalize your data in transfer learning.