Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Allow augmentations without DataKey.Input

See original GitHub issue

🚀 Feature

Sometimes you want to apply augmentations to e.g. masks the need to transform also an image.

Alternatives

None

Additional context

When using AugmentationSequential like this:

self.random_aug = AugmentationSequential(
   K.RandomVerticalFlip(),
  data_keys=["mask"]
)

following error is thrown: NotImplementedError: The first input must be DataKey.INPUT.

Issue Analytics

State:
Created 2 years ago
Reactions:2
Comments:7 (3 by maintainers)

Top GitHub Comments

1reaction

edgarribacommented, Dec 16, 2021

Agree, but for now they are treated the same, must be the user who specifies the interpolation method. I don’t think we support out of the box such fine-grained automation. In principle, for such cases we recommend to create custom classes and define different behaviors per data type.

1reaction

maaftcommented, Dec 16, 2021

Are they really currently treated the same for geometric augmentations? IMHO the interpolation method must differ.

One should never use other interpolation methods other than nearest for mask, or you’ll end up with non-binary values.

Same for actual metric data like e.g. depth-images