This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: Read Multiple layer output in TIDL Application

Part Number: TDA4VM

Hi,

I have a tidl application where I am running the dasnet model. 

Right now in all the tidl applications output of the last layer is used for postprocessing.

Suppose I want to get the information from multiple layers of the model to be used for post processing in my tidl od application, Is this possible to do so? 

Please suggest.

  • Hi 

    I hope you are referring below doc for using TIDL example application. If not share the link to the same.

    software-dl.ti.com/.../md_tidl_sample_test.html

    Use the below option in infer config file to select the post processing data ID

    postProcDataId Output Tensor ID on which the Post processing needs to be performed, 0 : Default
  • Hi,

    I will look into this parameter "postProcDataId".

    One thing that i see is that, in the link

    https://software-dl.ti.com/jacinto7/esd/processor-sdk-rtos-jacinto7/08_00_00_12/exports/docs/tidl_j7_08_00_00_10/ti_dl/docs/user_guide_html/md_tidl_sample_test.html

    Steps to run Sample Application on EVM

    The above inference test can be executed on Jacinto7 SoC using the TI_DEVICE_a72_test_dl_algo_host_rt.out binary (distributed with the installation). This section will describe the steps required to run the imported models on target.

    H/W requirements

    • TI Jacinto7 EVM
      • The EVM should be programmed to SD-boot mode as described in SDK user guide

    Preparing the SD card

    • Run the below commands in you linux machine to copy the imported models, input files and binary files required to the TIDL application on target
    user\@ubuntu-pc\$ cd ${PSDKRA_PATH}/vision_apps
    user\@ubuntu-pc\$ make linux_fs_install_sd



    These Steps only talks about the linux + rtos mode. Suppose i want to test the sample application
    on the EVM for RTOS+QNX mode, there is no TI_DEVICE_a72_test_dl_algo_host_rt.out binary available for the QNX.


    How do we test the standalone model on target, if i have QNX+RTOS boot.?

  • Hi,

    Please guide us on how do we test the standalone model on target, if i have QNX+RTOS boot.?

  • Hi,

    Regarding setting up the vision apps infrastructure for QNX + RTOS with SD boot, this can also be seen in the Vision Apps User Guide.   The steps for Linux and QNX are very similar for creating the required SD card content.

    Regarding generating the .out files, such as TI_DEVICE_A72_test_dl_algo_host_rt.out, for QNX, will need to check and get back to you. 

    Regards,

    kb

  • HI, 

    I don't have any problem in setting up the vision_Apps architecture for QNX+RTOS. i am able to create the applications in vision apps and run them.
    However i have issue with TIDL standalone testing, because the document only talks about the Linux mode while testing the tidl standalone application. 

    I want to know how can i test the standalone model on target, if i have QNX+RTOS boot mode on TDA4VM?

  • Hi Kumar,

    The TI_DEVICE_A72_test_dl_algo_host_rt.out, would need to be built for QNX, and run in the same fashion on target as is being for Linux.

    In regards to building the .out for QNX, the makefiles currently do not support this functionality but it should be possible to modify them to accomplish this,  similar to the build steps being done for Linux. 

    Regards,

    kb