29
votes

In Keras,

I'm trying to import _obtain_input_shape as follows:

from keras.applications.imagenet_utils import _obtain_input_shape

However, I get the following error:

ImportError: cannot import name '_obtain_input_shape'

The reason I'm trying to import _obtain_input_shape is so that I can determine the input shape(so as to load VGG-Face as follows :

I'm using it to determine the correct input shape of the input tensor as follow:

input_shape = _obtain_input_shape(input_shape,
                                  default_size=224,
                                  min_size=48,
                                  data_format=K.image_data_format(),
                                  require_flatten=include_top)`

Please assist? Thanks in advance.

7
There is probably a better way to do what you're trying. What exactly is it you want to do with _obtain_input_shape?Daniel Möller
Hi @DanielMöller, I have edited my question to include more information .Tshilidzi Mudau
You know that keras doesn't have a VGG-Face, right? The input shape should be chosen by you when you create a model. If you want the default input shape, just pass (224,224,3).Daniel Möller
Thanks for the responses Daniel, yes I know. I want to load VGG-face pre-trained weights. I will try what you suggested and let you know .Tshilidzi Mudau
Anybody know how to import this for tensorflow.keras?Austin

7 Answers

57
votes

You don't have to downgrade Keras 2.2.2.

In Keras 2.2.2 there is no _obtain_input_shape method in the keras.applications.imagenet_utils module. You can find it under keras-applications with the modul name keras_applications (underscore).

So you don't have to downgrade your Keras to 2.2.0 just change:

from keras.applications.imagenet_utils import _obtain_input_shape

to

from keras_applications.imagenet_utils import _obtain_input_shape
11
votes

I have found a method that works well. You just use

from keras_applications.imagenet_utils import _obtain_input_shape 

Notice: It is keras_applications instead of keras.application.

7
votes

This issue occured because of the version of keras.

In my case, I was downgrade keras 2.2.2 to 2.2.0, and the problem was solved.

1
votes

keras_applications.imagenet_utils is deprecated

Traceback (most recent call last): File "inception_v3.py", line 36, in from keras_applications.imagenet_utils import _obtain_input_shape ModuleNotFoundError: No module named 'keras_application

1
votes

In Colab I solved it by importing Keras and installing :

import keras
!pip install keras_applications
from keras_applications.imagenet_utils import _obtain_input_shape
1
votes
from keras.applications.imagenet_utils import obtain_input_shape

Not _obtain_input_shape. This works fine with keras==2.5.0rc0 (pip install keras==2.5.0rc0)

0
votes

for keras 2.2.4: Change the line like below to make it work.

from keras_applications.imagenet_utils import _obtain_input_shape

Note: It is importing from keras_applications and does not from keras.applications as before.