AutoAugment image augmentation about keras-cv HOT 9 CLOSED

I think that we could also evaluate differentiable augmentations like:
.....
https://openreview.net/forum?id=St-53J9ZARf

https://github.com/MSU-MLSys-Lab/DeepAA

from keras-cv.

I've received numerous inquiries about AutoAugment, so I definitely think we want this component.

from keras-cv.

Lets deprioritize in favor of RandAugment; which serves a similar purpose but is much easier to tune.

from keras-cv.

I've been weighing adding this layer.

Thoughts on what it adds over RandAugment?

It looks as though it's much more customizable, which could be good to have.

@innat

from keras-cv.

I think that we could also evaluate differentiable augmentations like:

https://arxiv.org/abs/2104.04282
https://arxiv.org/abs/2109.15273
https://openreview.net/forum?id=St-53J9ZARf

from keras-cv.

(AFAIK), the advantages of auto-augment over rand-augment also bring computational costs. I think both of them are used and maintained in research, so it's ok to have them both.

• AutoAugment - Code.
• RandAugment - Code
• Review.

from keras-cv.

I think that we could also evaluate differentiable augmentations like:
https://arxiv.org/abs/2104.04282
https://arxiv.org/abs/2109.15273
https://openreview.net/forum?id=St-53J9ZARf

/cc @haifeng-jin

from keras-cv.

this should probably use RandomAugmentationPipeline.

from keras-cv.

Has #900 was cloud I will unify the references on this ticket.

from keras-cv.