The pull request contains changes to the following files:
lib_data.py:
imgaug has been replaced with RandAugment from keras_cv, which relies on tensorflow v2.16.1
An additional check for the magnitude parameter in RandAugment is added, warning is raised if out of bounds (warnings will be generated if ignore is modified in future iterations)
sklearn.preprocessing dependencies are now up-to date
lib_model.py:
tensorflow_addons is now replaced with keras_cv and there are two possible loss functions (added in comments). The implemented FocalLoss() is the same as the focal loss used in tensorflow_addons (please click the link for more info)
The pull request contains changes to the following files:
lib_data.py:
lib_model.py:
Kind Regards, Prakash