AI Developer Cloud - Version Crash

[Duckpaddle]

Using https://stm32ai-cs.st.com , starting with my .tflite model. My goal is to incorporate the new model into the STM32N6-GettingStarted-ObjectDetection example.  I am having some issues with versions.

The ST Edge AI Developer Cloud produces a version of the model which differs from the version used by the support files in the example code. Case in Point - the function LL_ATON_LIB_Concat: the example has 6 parameters and code generated from the ST Edge AI requires 7 parameters.  

Dr Google, suggests there is another option when I "Download C Code" from the AI Developer Cloud that I should also have an option to change the Target Output to "STM32CubeIDE Project", "System Performance Project", or "C-Project".

I don't see that option. can you help.

[Julian E.]

Hi @Duckpaddle,

 

First of all, make sure to use the same version of the st edge ai core as the one in the getting started you use.

If you have the latest version of the getting started, this should be 4.0, which should align with what you used in the dev cloud.

 

The getting started is made to deploy models from the ST Model ZOO, as is, or retrained, I don't know what model you use, but you may have modifications to do.

In any case, the way to deploy models is explained here:

STM32N6-GettingStarted-ObjectDetection/Doc/Deploy-your-Quantized-Model.md at main · STMicroelectronics/STM32N6-GettingStarted-ObjectDetection · GitHub

The difference you saw in the LL_ATON may be due to compiler option specific to the profile that you will use by following the doc (so it is different than the default option in dev cloud, ie, the model generated is not exactly the same).

 

Lastly, there is indeed multiple option at the end of the dev cloud for most stm32, but not for the N6 and probably not for other MCUs with accelerated hardware (example u3 hsp).

 

Maybe, an option for you is to use the STM32CubeAI Studio to generate a simple template application that runs your model on the board and start from there to add the camera and display etc, if you encounter issue by adapting the getting started (which is already a complicated application)

 

Have a good day,

Julian

 

[Duckpaddle]

Thank you for the current version number, that helps a lot.

I have built a YOLO model for industrial control, not in the zoo but in the wild. 

I attempted to use the cli stedgeai version 3.

 

Thank you,

George

[Julian E.]

Hi @Duckpaddle,

 

The new ST Model Zoo added remade a tutorial on how to retrain yolov8n, yolov11n and yolo26.

This may be helpful for you.

The model zoo supports their deployment also:

ultralytics/examples/YOLOv8-STEdgeAI/README.md at main · stm32-hotspot/ultralytics · GitHub

 

Have a good day,

Julian