Validity of Augmenting augmented images in training dataset. (Medical Data)

Viewed 13

Training Images : 50

Augmented Images: 15000

Normally, I see papers using stand-alone augmentations. But I tried combining them. Example: Rotating -> Translating -> Zooming -> Sharpening.

I got excellent results on the test set.

But I feel the model is memorizing rather than learning.

Question: Could I have some opinions on this? Am I doing the right thing?

0 Answers
Related