This training module has a BUG.

[wotchin]

The file "utils/data_augmentation.py" has a bug,so if you want to train the model(I only tested the gender part),you will get an exception:

ValueError: No data provided for "image_array_input". Need data for each key in: ['image_array_input']

How to solve this issue? I changed the code in the "data_augmentation.py" file:

change

    def _wrap_in_dictionary(self, image_array, targets):
        return [{'input_1':image_array},
{'predictions':targets}]

to

    def _wrap_in_dictionary(self, image_array, targets):
        return [{'image_array_input':image_array},
                {'predictions':targets}]