U-Net models no longer work with recent versions of Tensorflow

[Pedro-Filipe]

Can we update the model files to use recent versions of TensorFlow?

At the moment I get this error:

Exception encountered: Unrecognized keyword arguments passed to Conv2DTranspose: {'groups': 1}

I think the problem is this:

I can now confirm that if the model is created in Keras 3, and saved as .h5 the problem does not occur. It only occurs with historic models created in tf.keras. However, we (and customers) have a large collections of such models.

link to webpage with issue

[tito21]

If you still have the data sorted somewhere, it might be worth retraining with nnUNet. The project seems well maintained and easy to use for inference.
Plus it would be an easy abstract 😄