Tensorflow Hub Inception V3 structure compared to Keras Inception V3 structure?

问题:

I have been working Keras Inception V3. I believe the Keras model structure looks like the image below. But when I use the Inception V3 model from Tensorflow Hub, I think (but am not sure) that the TF-Hub Inception V3 model includes up until the Mixed7 output of IV3 (the 7th red "Concat" box), where Keras, in contrast, includes up to the Mixed10 output (10th red "Concat" box). Can someone confirm? (or deny!) the structure of the Tensorflow Hub Inception V3 Feature Vector model? Can someone tell me where to find Tensorflow Hub documentation showing images/diagrams of TF-Hub models to clarify?

回答1:

The naming conventions differ between Keras and SLIM, unfortunately. https://tfhub.dev/google/imagenet/inception_v3/feature_vector/1 was created from SLIM code and uses its naming conventions with numbers and letters, see http://github.com/tensorflow/models/blob/master/research/slim/nets/ (or explore the Hub module on TensorBoard while training, if you are using it).

文章来源: Tensorflow Hub Inception V3 structure compared to Keras Inception V3 structure?