Generate file .ucf with NanoEdgeAI

[ChristianValota]

Good afternoon,

I want to upload a Machine Learning algorithm on the ISPU part of the LSM6DSO16IS. My alghoritm has to identify is the subject is stationary, walking or running. I used NanoEdgeAI Studio to develop the classification algorithm. By doing this I obtain these files, but don't the .ucf file, which is the file that I have to upload on the ISPU part of the sensor. By uploading this files on the Nucleo, it performs the classification correctly, but I want the classification to be done by the sensor. Which is the easyest way to obtain it starting from the files that I obtained? 

 

[Julian E.]

Hello @ChristianValota,

 

It seems you didn't select the ISPU as target in your project settings.

The ISPU is only available for anomaly detection in NanoEdge under the name ISM330ISN.

So, I don't think you can do your multiple classes classification...

You can look at this tutorial if you decide to do anomaly detection with the ISPU to understand how to deploy a model: https://wiki.st.com/stm32mcu/wiki/AI:NanoEdge_AI_Anomaly_Detection_library_for_ISPU 

 

You may take a look to the Tool MEMS Studio instead: MEMS-Studio - Software solution for MEMS sensors with graphical no-code design of algorithms and development of embedded AI features - STMicroelectronics

(I never used it, so I am not sure)

 

Have a good day,

Julian