My export to tensorflow2 with weights is not working as expected, help required

My export to tensorflow2 with weights is not working as expected, help required

In the confusion matrix, everything appears in a single column - i.e all the weights have been set to 1

[dpb]

which -all tensorflow2

'tensorflow2' not found.

Doesn't seem to be a Mathworks-supplied function; so what is it and how are we supposed to have any idea what might have gone wrong with absolutely nothing to look at???

[Dr Peter Hobden]

Deep learning toolbox

Command: ExportNetworkToTensorflow

[Dr Peter Hobden]

Exports okay, but when you create the network in python by running the matlab generated script, the weights look to be corrupted.

[dpb]

Minimum working example that illustrates the issue?

[Dr Peter Hobden]

Export a resnet model

Only changes in my example are that the input shape is 228,227 3 and classes are 4

import resnet50python

model = resnet50python.load_model()

#* To save a loaded model into TensorFlow HDF5 format:

model.save('resnet50.h5', save_format='h5')

def create_model():

input_1_unnormalized = keras.Input(shape=(227,227,3), name="input_1_unnormalized")

input_1 = keras.layers.Normalization(axis=(1,2,3), name="input_1_")(input_1_unnormalized)

...