1
$\begingroup$

I am trying to import Inception v3 in TensorFlow. I wish to apply it after reading this tutorial on object detection.

$\endgroup$
2
$\begingroup$

Keras, now fully merged with the new TensorFlow 2.0, allows you to call a long list of pre-trained models. If you want to create an Inception V3, you do:

from tensorflow.keras.applications import InceptionV3

That InceptionV3 you just imported is not a model itself, it's a class. You now need to instantiate an InceptionV3 object, with:

my_model = InceptionV3()

at this point, my_model is a Keras Sequential() model with the architecture and trained weights of Inception V3, that you can re-train, freeze, save, and load as you need.

Check also the full list of models available in the module, it's great.

$\endgroup$
  • $\begingroup$ Does this apply with docker as well $\endgroup$ – Blackreaved Nov 29 '19 at 12:20
  • $\begingroup$ Unfortunately I don't have practical experience with Docker. However, since the GCP is configured for using containers, I think it can be applied without problems $\endgroup$ – Leevo Nov 29 '19 at 13:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.