Data Augmentation
Data Augmentation is a technique used to “increase” the amount of data to train a model.
This can be helpful in cases such as when you don’t have a sufficient amount of data or when you want to increase the diversity of your training set.
What it actually does
When applying data augmentation, it doesn’t actually “increase” the number of samples of your training data.
What happens is that by randomly applying augmentations every time a data sample is retrieved, the model won’t see the exact same image which improves the generalization and prevents overfitting.
