You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think PIL implementation assumes uint8 dtype, and TF's bitwise ops only support integers.
Is it safe to assume that this layer's input will only accept integer values in range 0-255? Or should it accept any dtype, cast internally to uint8, posterize and cast back to original dtype?
https://github.com/tensorflow/similarity/blob/906d141b75acf00d0e1d37dd0d88c432bbb30f57/tensorflow_similarity/augmenters/img_augments.py#L494
The text was updated successfully, but these errors were encountered: