weights: one of `None` or 'imagenet'

[ganab...@gmail.com]

Hello

I have done with training and got my model(weight). Now i want to use that model for extracting features (or prediction)

But the architectures from Keras (such as vgg, xception,...) only allow to use imagenet pre-trained model.

Must i create my own architectures to use my pre-trained models?

Or keras still provides source that allows us to use our own weight? if yes where is that source and how to use it?

Error: 

if weights not in {'imagenet', None}:
	raise ValueError('The `weights` argument should be either '
	'`None` (random initialization) or `imagenet` '
	'(pre-training on ImageNet).')

My code:

weights='./data/checkpoints/Xception-.001-0.588_2.hdf5'

base_model = Xception(

                weights=weights,

                include_top=True

                )

[Matias Valdenegro]

If you have a trained model, you can use it directly for feature extraction by
building a function that computer a given layer output given the input. which
is covered in the Keras FAQ.

I am not sure why you introduce the pre-trained imagenet models, they
literally have nothing to do with what you are trying to do.

[ganab...@gmail.com]

Thank you for your reply.

The code above is how to use pre-trained weight. 

But the architectures only allow imagenet weight in keras code. this is what i meant.

Anyway succeeded,.. i had to build all architectures by myself to use my own weight. 

[SHEFALI]

How did you solve this problem? Can you explain it briefly

[Alok Das]

weights='./data/checkpoints/Xception-.001-0.588_2.hdf5'

base_model =  weights  (

                weights="none",

                include_top=True

                )