2020-07-09 07:17 AM
Dear STM community,
I am trying to deploy a simple tflite model on my STM32, but I can't get through the analyze step. I got the following error:
TOOL ERROR: Unknown layer type SHAPE, stopping.
If I convert the model to Keras (model and topology), it works fine, I can run the generated code, but I want to use tflite model.
I attached the keras and the tflite files.
Thanks,
Titusz
Solved! Go to Solution.
2020-08-05 01:36 AM
You have a SHAPE layer in your model, which is not currently supported in X-CUBE-AI. Shape layers extract the shape of tensors, so check your model if you use any .shape operation when building your model graph, and you can use Netron to exam your model graph.
You can look up supported layer types in official documents located in %USERPROFILE%/STM32Cube/Repository/Packs/STMicroelectronics/X-CUBE-AI/5.1.2/Documentation/layer-support.html
2020-08-05 01:36 AM
You have a SHAPE layer in your model, which is not currently supported in X-CUBE-AI. Shape layers extract the shape of tensors, so check your model if you use any .shape operation when building your model graph, and you can use Netron to exam your model graph.
You can look up supported layer types in official documents located in %USERPROFILE%/STM32Cube/Repository/Packs/STMicroelectronics/X-CUBE-AI/5.1.2/Documentation/layer-support.html
2020-08-10 01:28 AM
Thank you for your answer, it helps me a lot! ;)
2021-03-24 01:10 AM
I was facing the same error, how did you get around it? Is there any alternative to .shape?
2021-03-24 01:26 AM
I'm so sorry, I can't help you, because it wasn't a top priority project and we stopped working on it.
2021-03-24 01:35 AM
Sure, thanks for the reply.