deploy ai model on the STM32H743

[llcc]

I want to deploy an ai custom model on the STM32H743. Are there any relevant tutorials or examples?

[mƎALLEm]

Hello,

Is that what you are asking for: Deploying a deep learning model on NUCLEO-H753ZI? 

[Julian E.]

Hi @llcc,

 

For STM32H743, the process to deploy a model is standard, so it is not as difficult as for the N6.

The only thing to do is to download the latest version of X Cube AI in CubeMX.

Then create a project, activate X Cube AI, select the application template.

Browse for your model in the X Cube AI interface, run a analyze to make sure that your model is being compiled correctly, then generate the code with cubeMX and that's it.

 

Have a good day,

Julian

[llcc]

Hi, Does the model deployed on the STM32H743 need to be quantized first like the n6, or can floating-point types be deployed directly? Additionally: After the deployment is completed, how should I pass the data to the model for inference and obtain the inference results? If using serial communication for data transmission, do I need to additionally configure the serial port and clock in CubeMX? Is there example code for specifically passing input samples via serial communication? Thanks!

[Stone_chan]

HI , llcc

Please check settings , some variables are not automatically generated , like IO buffers variables

 

 

 

 

[llcc]

Thank you! The problem has been resolved; it appears to have been an issue with the clock configuration.