• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 98 subscribers
  • Views 353 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.

SK-TDA4VM: Meaning of parameter

Ryuuichi machida
Guru 13385 points

Part Number: SK-TDA4VM

Hello,

I would like you to ask following paramter.

* When I performed AI demo, I got following log on console.

Then, I have following question.
Q1, What does "samples" mean ?
Are Inference, total time and frame rate parameters averaged ?
(I thought inference will be performed for each frame, so these parameters should be output as just "time".)

BR,

  • 0
    TI__Guru 61153 points

    Hi Machida-san,

    "samples" mean total frames that have been processed. The numbers make more sense when running the demo with the input source set to /opt/edge_ai_apps/data/images/%04d.jpg which handles a single image every second.

    Total time and framerate are averaged, but dl-inference time is not averaged and reports the time taken for running the model on each frame.

    Regards,

    Takuma

  • 0 in reply to Takuma Fujiwara
    Guru 13385 points

    Hello Fujiwara-san,

    "samples" mean total frames that have been processed. The numbers make more sense when running the demo with the input source set to /opt/edge_ai_apps/data/images/%04d.jpg which handles a single image every second.

    I could understand by running above demo.
    sample number is incremented by 1. 

    >Total time and framerate are averaged, but dl-inference time is not averaged and reports the time taken for running the model on each frame.
    Let me clarify about above meaning.
    Ex, I observed following log at first.
    | dl-inference: 5.87 ms from 17 samples |
    | total time: 42.18 ms from 15 samples |
    | framerate: 23.71 fps from 15 samples |

    In that case, is my following understanding correct ?

    For "dl-inference" : "5.87ms" means only showing inference time at 17th flame.
    For "total time", "framerate" : It is the average value until 15 flame
    => if so, can we see log file which show operationg time for each flame ?

    BR,

  • +1 in reply to Ryuuichi machida
    TI__Guru 61153 points

    Hi Machida-san,

    Unfortunately, no logs are generated for showing operating time for each individual frame. If there is a need, then please implement this by reference our code.

    What we do for total time is we measure the time taken from when the demo starts, and framerate is calculated by dividing total frames processed by total time. Please refer to edge_ai_apps/utils.py's Report class for the code for calculating total time and framerate.

    Regards,

    Takuma

  • 0 in reply to Takuma Fujiwara
    Guru 13385 points

    Fujiwara-san,

    Understood.

    BR,