INTERNAL ERROR: Inconsistent in/out tensor shapes

ahmedbing · ‎2024-03-18

Hello, I tried to convert my model using STM32Cube.AI Developer Cloud and on my pc STM32Cube AI at CubeMx. However, I got following error from stm32ai. Then I debug model with netron and realize that sizes are not matching with stm32ai library. How can i solve that issue?

The layer image is

And output from STM32Cube.AI Developer:

>>> stm32ai analyze --model nanodet.onnx --allocate-inputs --allocate-outputs --compression none --optimization balanced --series stm32f4 --target stm32f4 --name network --workspace workspace --output output
Neural Network Tools for STM32 family v1.7.0 (stm.ai v8.1.0-19520)
INTERNAL ERROR: Inconsistent in/out tensor shapes in _head_Reshape_output_0_to_chlast_output: ['H: 32', 'W: 52', 'CH: 52'] and ['H: 33', 'CH: 2622'].

fauvarque.daniel · ‎2024-07-17

The model is now passing on the developer cloud.

You can use the command line STEdgeAI Core 1.0 and the model will pass

We will be soon releasing STM32Cube.AI 9.1 that will use STEdgeAI core 1.0 for NN conversion

In order to give better visibility on the answered topics, please click on 'Accept as Solution' on the reply which solved your issue or answered your question.