RZ/V2L DRPAI start issue

Dear Renesas team,

We have built yocto recipe for RZ/V2L and added drpai, ISP layer according to the document, found the drpi application sample and built/cross compiled  the application as per the below attached documents, r11an0573ej0700-rzv2l-drpai-application-note.pdf

but we are facing issue while running the app, please find the error in the  below image 

help us to sort out this issue.

Thanks & Regards

Mahesh R 

Parents Reply Children
  • Dear Michael kosinski,

    Have you loaded the latest bootloaders from your Yocto Build

    no we have not used

    The Bootloaders that are preloaded with the RZV2L board do not work with latest BSP drivers including DRP and ISP.

    will update the boot-loader and will let you know the status ASAP.

    Thanks & Regards

    Mahesh R

  • Dear Michael Kosinski,

    As per your guidelines we have updated the boot-loader and now we were able to run the DRP-AI example code, we also cross complied the drp example code and it was also successful, we have followed this document and which tell that the output result will also push to hdmi monitor, but for us, we are not able to see result in the HDMI monitor, only weyland screen is visible,

    please find the attached reference doc and log images 1586.r11an0573ej0700-rzv2l-drpai-application-note.pdf

       please help us to see result in HDMI monitor 

    Thanks & Regards

    Mahesh R

  • Hi Mahesh,

    we're glad to hear that the DRP-AI starts-up properly now.

    The "*_img" sample applications do not show something on the display.

    Please try

    - app_tinyyolov2_isp with Coral camera or

    - app_resnet50_cam with USB camera

    Regards,

    Stefan

  • Dear Stefan,

    Thanks for the support, we were able to run the demo applications and explore the drpai and isp feature in RZ/V2L.

    will let you know if any help is needed.

    Thanks & regards

    Mahesh R

  • Dear Renesas Team,

    Does DRP-Ai Translator_RZV2L supports MobileNet_SSD.onnx Model ?

    because when we run run_DRP_AI_translator_V2L.sh script, we could see the list of supported models are:-

    1)Resnet50V1

    2)VGG16

    3)tiny_yolov2

    4)yolov2

    Thanks & Regards

    Mahesh R

  • Dear Mahesh,

    The neural networks that can be automatically translated with run_DRP_AI_translator_*.sh are just examples. 

    By using script arguments, you can translate your own .onnx  easily.
    I'm using the following commands:

    mkdir -p ./output/${CASE}
    ./run_DRP-AI_translator_${TARGET}.sh ${CASE} \
    -onnx onnx/${ONNX_MODEL}.onnx \
    -addr UserConfig/addrmap_in_${CASE}.yaml \
    -prepost UserConfig/prepost_${CASE}.yaml \
    -CheckPrePost ON \
    -SaveOptOnnx \
    >&./output/${CASE_STD}/DRP_AI_Translator.log

    Mobilenet can be translated as well. Please check the "rzv2l_ai-implementation-guide". You'll find a description for Pytorch Mobilenet.

    MobileNet_SSD is a bit different. It cannot be translated by the DRP-AI Translator yet. We recommend to switch to use TinyYOLOv2/v3 instead.

    Please note that DRP-AI translator enhancements are under development continuously. With each release, more deep neural networks can be translated.

    Regards,

    Stefan