def data_augmenter():
data_augmentation = tf.keras.Sequential([RandomFlip("horizontal"),RandomRotation(0.2)])
return data_augmentation
I used the above code for the augmentation but it worked just once on the image. When I tried for the second and third times it wasn’t applying any augmentation on the image. Whats the reason behind it???
All you have done there is define a separate Keras sequential model.
Models are used for training - they aren’t stand-alone functions.
When I used the code data_augmentator()(train_x[i], training =true) the code works fine and It applies the augmentation. But when the code has data_augmentator()(train_x[i]) it doesn’t work.
Which week number and assignment are you working on? You didn’t include that info in your post.
Course 4 week 2 assignment of Mobile net V2.
Thanks, I’ve updated the thread title. I’ll check into your issue shortly.
You do not need to add any code that uses data_augmenter() directly.
Its use is limited to this shortcut in the function definition:
You should not modify that line of code.
Within the alpaca_model() function, you only use data_augmentation() on the “inputs” variable, like it says here in the code comments:
# apply data augmentation to the inputs
That’s not the correct usage. And, “data_augmentator()” does not exist. The function is “data_augmentation()”.
And it doesn’t need the “training” parameter.