Analyze is failing and got error as NOT IMPLEMENTED: Transpose of batch not supported Got the above error while analyzing the tfltie model in STMCubeMX

[Umahe.1]

I have an AI model that detects the human activity and saved the model in .h5 format and converted it into .tflite image. When I upload the image into STMCubeMX and analyze the model, analyze is failing and showing as NOT IMPLEMENTED: Transpose of batch not supported.

Need help with the resolution for above issue. Thanks in Advance.

[fauvarque.daniel]

Does the Keras model pass in X-CUBE-AI ?

Could you share the tflite model so we could have a look

[Umahe.1]

Thanks for the response and have attached the model.

[fauvarque.daniel]

I can reproduce your error with your model, thanks.

I see that your network is in float, did you try to directly import the h5 Keras model ? as X-Cube-AI natively supports Keras it is better to take the original model rather than a transformed one

Regards

[Umahe.1]

I'm Okay if my model predicts and throws output in float, I just want the model to work with MCU.

Tried with h5 model and seeing this error:

INTERNAL ERROR: The layer has never been called and thus has no defined input shape.