STM32CubeMX, X-CUBE-AI error occurring from analysis of prebuilt .tflite network [AI:network] NOT IMPLEMENTED: Unsupported code generation for layer 'Detection' [AI:network] Analyze fail on the AI model

[GWall.2]

Hi all,

Currently using the STM32H747I-DISCO with camera to explore using computer vision for people detection. I have already successfully used one of the example NN (found in the FP-VISION-AI1) and now look to use the X-Cube-AI to integrate further tensorflowlite NNs.

I'm working off of the Artificial Intelligence (AI) and computer vision function pack​ for STM32H7 microcontrollers - User manual and it states:

As such I have loaded CubeMX for the STM32H743ZITx to allow Cube-AI functionality, however, when loading various official tensorflowlite files, I always get the same error message when analysing the model:

The .tflite files can be found at: https://tfhub.dev/s?deployment-format=lite&module-type=image-object-detection

EDIT: could this issue be with the TFLite_Detection_PostProcess node as shown below in netron:

 

Any help would be much appreciated.

Many thanks,

Greg

[GWall.2]