Use models from Keras Applications without pretrained weights

Question

Keras Applications provide implementations of some of the most popular model architectures with weights pretrained on some of the most popular datasets. These predefined models are very handy for transfer learning of problems which are similar to the datasets the models were trained on.

But what if I have a very different problem and want to completely train the models on the new dataset? How can I use the models in Applications for training from scratch based on my own dataset, if I dont have pretrained weights?

Simbarashe Timothy Motsi · Accepted Answer · 2019-11-16T17:51:48.437

5

You can assign a None to the weights variable, for instance with the inception V3 architecture.

keras.applications.inception_v3.InceptionV3(include_top=False, weights='None', input_shape=input_shape = (img_width, img_height, 3))

include_top=False will allow you to train the top layer with your custom network.

weights='None' means that we are training without any weights if you want to train using imagenet weight you set it to weights='imagenet'

edited Nov 16 '19 at 17:51

answered Jun 04 '18 at 08:21

Simbarashe Timothy Motsi

1,469
14
28

I.e. I use include_top=False to replace the top layer by my own top layer (based on my number of classes) and set the weights to 'None' to train the complete model from scratch? – Oblomov Jun 04 '18 at 08:37
2

I think it should be None not 'None' – Sadaf Shafi Mar 04 '21 at 13:44
it should be None not 'None' – ahmedshahriar Feb 17 '22 at 14:29

Use models from Keras Applications without pretrained weights

1 Answers1