Changing batch input shape about keras-vgg16-places365

[GKalliatakis]

Code has now been updated and is fully keras-applications-compatible, which means the models can be utilised like any default keras models (taken from keras.applications).

Please try again with the updated code and comment below.

[fmaglia]

Thanks!
I tried the 2 scripts and now I have another problem (both whit VGG16_Places365 and with VGG16_Hubryd_1365):

File "Dottorato/vgg16_hybrid_places_1365.py", line 101, in VGG16_Hubrid_1365
    include_top=False)
TypeError: _obtain_input_shape() got an unexpected keyword argument 'include_top'.

Finally is there a possibility to change the input shape of a network for every image?
Or can I release the model from the GPU memory for at the next iteration regenerating the same network model with a different input shape?

[GKalliatakis]

Check that you are up-to-date with the master branch of Keras. You can update with:

pip install git+git://github.com/fchollet/keras.git --upgrade --no-deps

Regarding input shape please refer to the official documentation :

input_shape: optional shape tuple, only to be specified if include_top is False (otherwise the input shape has to be (224, 224, 3) (with 'channels_last' data format) or (3, 224, 224) (with 'channels_first' data format). It should have exactly 3 inputs channels, and width and height should be no smaller than 48. E.g. (200, 200, 3) would be one valid value.

[fmaglia]

import keras
Using TensorFlow backend.
print(keras.version)
2.1.1

I'm using the latest version!

[GKalliatakis]

Have a look at the following:

• https://yohanes.gultom.me/keras-vgg16-with-different-input-shape/
• https://stackoverflow.com/questions/42187425/how-to-change-input-shape-in-sequential-model-in-keras

[GKalliatakis]

Also for

TypeError: _obtain_input_shape() got an unexpected keyword argument 'include_top'.

Try changing "include_top" to "require_flatten" (answer taken from rcmalli/keras-squeezenet#13).

[fmaglia]

Thanks.
Now it works!