• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 258 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

IWRL6432: IWRL6432 demo time consumption query

Ken Zhang
Expert 3395 points
Part Number: IWRL6432

Hi Experts.

Do we have time consumption bench mark for IWRL6432 People counting demo?

Typically how long will processing chain/tracking/classification take to finish processing for each of these blocks?

Thanks

Ken

  • 0
    TI__Mastermind 37785 points

    Hi Ken,

    We actually print this out in the stats TLV. You can enable this in the guiMonitor CLI command. 

    Best,

    Nate

  • +1
    TI__Intellectual 790 points

    Hello Ken.

    When the classification module is disabled (in the cfg file as explained in the tuning guide) and the default chirp configs in the SDK are used, the entire processing chain (detection and tracker) takes ~30-35ms for 3 tracks (depending on the size of point cloud, etc.). Each additional track needs ~0.5ms additional processing time.

    When the classification is enabled:

    - For the 2-layers CNN model, we expect ~3ms additional processing time per track

    - For the 3-layers CNN model, we expect ~15ms additional processing time per track.

    Here, 2-layers or 3-layers CNN model can be configured in compile time.

    Regards.

    Muhammet 

  • 0 in reply to Muhammet Yanik
    TI__Expert 3395 points

    Hi Muhammet

    Thanks for providing the detailed benchmark.

    Ken