Showing results for 
Search instead for 
Did you mean: 

Number of Validation Inputs for STM32CubeMX

Associate II

I am using the STMicroelectronics.X-CUBE-AI.8.0.1 with the board STM32U5A5ZJTxQ 


Here is what I did:

  1. Applauded a keras model under the form .h5. 
  2. Putting PD9 and PD8 on the USART3_RX and USART3_TX respectively 
  3. Using the connectivity to USART3 in mode Asynchronous and Hardware Flow Control disable
  4. Systel Core RCC:
    1. Put HSE to BYPASS Clock Source
    2. Put LSE to BYPASS Clock Source
  5. Clock Configuration: 
    1. MMary11_1-1689680658365.png
  6. Activate the CRC
  7. Software Packs Component Selector
  8. Put STMicroelectronics.X-CUBE-AI Application to validation
  9. Activate the AI X-CUBE-AI and Device Application
  10. Put the Platform Settings to COM Port: USART:Asynchronous USART3

Question: at the network place: is it possible to add two datasets, so two files, in the inputs into the validation inputs section or/and in the validation outputs? If yes, is it possible to do it for several datasets?


Thank you in advance,


Mahmoud Ben Romdhane
ST Employee

Hello @MMary.11 

First of all, I want to thank you for reporting.
STMicroelectronics X-CUBE-AI focused on converting and optimizing the pre-trained neural network models.

The training process is done by using external deep learning frameworks like TensorFlow or PyTorch on the Host Machine.
The validation process is a part of the training stage, and it is performed outside STM32Cube AI. That's why you can use many validation datasets depending on the complexity of your model and the size of your datasets.
I advise you to refer to the official documentation and resources provided by STMicroelectronics.

AI:X-CUBE-AI documentation - stm32mcu