Unwanted behavior when not loading a transformer in CDON dataset

[alessioGalatolo]

Supervised-Image-Classification-with-Noisy-Labels-Using-Deep-Learning/Custom_dataset.py

Line 24 in 19a3be1

if self.transform:

If no transformer is present the image object will be close before the return and unexpected behavior can happen.

Priority: very very low (we never call it without a transformer)