How to make a model differentiate between two visually similar classes?

vision
1

I have two classes that are extremely similar, in fact, my model predicts 99% of class 1 as class 0.

Someone recently recommended to enhance the differences between the two classes by applying some transformations to one of the class. I’m not too familiar with the process so I was wondering if it would work and if so, what should I change, Brightness? Exposure?

2

This is actually a topic in computer vision research called “fine-grained image classification.” You may take a look at the literature and pick a most relevant one. I personally recall some methods like “tripplet loss.” You may want to take a look at it.