Skip to main content
Snara.21
Senior
July 26, 2021
Question

How to deploy the tensorflow model in the Sensor tile kit.. Is there is any sample code for us to learn and understand how it works

  • July 26, 2021
  • 5 replies
  • 1291 views

..

    This topic has been closed for replies.

    5 replies

    Technical Moderator
    July 26, 2021

    Hello @Snara.2​ ,

    Please have a look at this ST wiki article and video, that may help you.

    You can refer to the Getting Started with TensorFlow (via Google search) that shows you how to use step by step the TensorFlow Lite for Microcontrollers.

    Imen

    In order to give better visibility on the answered topics, please click on 'Best answer' on the reply which solved your issue or answered your question. Thanks
    Snara.21
    Snara.21Author
    Senior
    July 26, 2021

    Hi @Imen DAHMEN​ 

    Thanks for your immediate response. I have followed the instruction shown in "dm00570145-getting-started-with-xcubeai-expansion-package-for-artificial-intelligence-ai-stmicroelectronics" . I have a Tensorflow lite model file, when I try to upload the model. Analyse is working fine, But validation on target . It says StCube32IDE not found

    Snara.21
    Snara.21Author
    Senior
    July 27, 2021

    @Imen DAHMEN​ 

    Starting AI validation on target with random data... 

    C:/Users/Admin/STM32Cube/Repository/Packs/STMicroelectronics/X-CUBE-AI/7.0.0/Utilities/windows/stm32ai validate --name network_1627366977421 -m E:/Firmstart/Comtustec/Client_Software/model_int8.tflite --type tflite --compression 1 --verbosity 1 --workspace C:\Users\Admin\AppData\Local\Temp\mxAI_workspace1105783762988500295032472845520054 --output C:\Users\Admin\.stm32cubemx --mode stm32 --desc 115200  

    Neural Network Tools for STM32AI v1.5.1 (STM.ai v7.0.0-RC8) 

    WARNING: nl_9 (SOFTMAX) in SIGNED not supported. Falling back to float 

    Setting validation data... 

     generating random data, size=10, seed=42, range=default 

     I[1]: (10, 30, 32, 1)/int8, min/max=[-128, 127], mean/std=[-1.960, 73.342], scale=0.07154346 zp=17, conv2d_1_input 

     No output/reference samples are provided 

    Running the STM AI c-model (AI RUNNER)...(name=network_1627366977421, mode=stm32) 

    E200(ValidationError): stm32: Unable to bind the STM AI runtime with "network_1627366977421" c-model: [] 

     connection to "serial:115200"/"115200" run-time fails 

     E801(HwIOError): Invalid firmware - COM4:115200 

    I need help to resolve this issue . ia m using Sensortile.box

    Technical Moderator
    July 28, 2021

    Hello @Snara.2​ ,

    I think this is known issue with X-Cube-AI package.

    I will check the status of this issue with the involved people and I'll make sure to post any updates here as soon as they come bak from summer vacation.

    I count on your patience and on your kind understanding.

    Thanks

    Imen

    In order to give better visibility on the answered topics, please click on 'Best answer' on the reply which solved your issue or answered your question. Thanks
    Snara.21
    Snara.21Author
    Senior
    August 9, 2021

    @Imen DAHMEN​  Hi Any inputs for this issue, this issue comes irrespective of the boards, I validated in STELVAL MKSBOX1V1 and STEVAL - STLK01V1. In MKSBOX1V! the STLINK1 is connected to PA9 and PA10 USART1_RX and TX and I checked the connection using multimeter. Is it possible to support as I need to validate and test a few deliverables