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: TI Auto Valet Parking Demo - Issues

Part Number: TDA4VM

Hi,

we started looking at some of the TI Auto Valet Parking Demos, but had some issues.

As the EVM boards are still not stocked we started to look at things which should run in emulation mode.

 Auto Valet Parking Demo:

https://software-dl.ti.com/jacinto7/esd/processor-sdk-rtos-jacinto7/07_02_00_06/exports/docs/vision_apps/docs/user_guide/group_apps_dl_demos_app_tidl_avp.html

 

  • no video output, only single images as files
  • wrong colors on the output images
  • very low performances on emulation mode

Auto Valet Parking Demo 3:

https://software-dl.ti.com/jacinto7/esd/processor-sdk-rtos-jacinto7/07_02_00_06/exports/docs/vision_apps/docs/user_guide/group_apps_dl_demos_app_tidl_avp3.html

did not run at all and gave following error:

 

0.0s:  VX_ZONE_INIT:Enabled

0.72s:  VX_ZONE_ERROR:Enabled

0.75s:  VX_ZONE_WARNING:Enabled

Computing checksum at 0x00007F83F5DF6D50, size = 262136

Computing checksum at 0x00007F83F58D4B90, size = 615560

0.371722s:  VX_ZONE_ERROR:[tivxAddKernelDmpacDofValidate:614] Temporal predictor is ON and pipeline is OFF. The flow vector input is expected to be handled by the caller and hence the 'flow_vector_internal_delay_num' should be 0.

0.371786s:  VX_ZONE_ERROR:[ownGraphNodeKernelValidate:531] node kernel validate failed for kernel com.ti.hwa.dmpac_dof at index 7

0.371790s:  VX_ZONE_ERROR:[vxVerifyGraph:1930] Node kernel Validate failed

0.371791s:  VX_ZONE_ERROR:[vxVerifyGraph:2098] Graph verify failed

MSC: Node send command failed!

 

Do you have some hints on how to solve here?

Regards,

Jon

  • Jon,

    By design the host emulation run is expected to,

    1. Run filed based input and file based output back-to-back with NO display

    2. Is NOT optimised for performance.

    3. Input and output images are in NV12 format

    The output image attached has to be viewed as an NV12 format (aka YUV420SP format). If you select the right color format in your viewer then you will see the correct output.

    For AVP3 demo, it uses DOF (Dense Optical Flow) and we need to set, enable_temporal_predicton_flow_vector 0, which means it uses spatial predictors. Temporal predictors is not required as it complicates data flow and provides no significant advantage over spatial predictors. This option will be removed in future releases.


    Regards,
    Shyam

  • thanks for clarifying this.

    regards,

    Stefan