CODEWITHSUNDEEP
tensorflowkeras
Matthias
Matthias

Reputation: 10379

What is the difference between TF model garden and tf.keras.applications?

With the new TensorFlow 2 we have their Model Garden (in GitHub under /models), as well as the pre-trained models for Keras under tf.keras.applications.

What is the difference between those two?

Upvotes: 1

Views: 375

Answers (1)

TF_Support
TF_Support

Reputation: 1836

Tensorflow Keras Applications
- pre-trained models for CNNs. They include most frequently used CNN architectures such as ResNet, InceptionNet, VGG etc.
- tf.keras.applications allow you to directly import a CNN architecture (see docs https://www.tensorflow.org/api_docs/python/tf/keras/applications?hl=en) customize its loss and use it as you like.

Tensorflow Model Garden
- models for other tasks such as NLP, object detection etc. These models are too maintained by TensorFlow.
- tf model garden are not part of tf core API so you can't import them directly as you do for tf.applications.

Upvotes: 1

Related Questions