Search code examples
object-detectionframe-rateyolodarknet

How to calculate various model speed in FPS for my GPU? In Darknet


I am using Darknet in my object-detection project. And I want to create a similar chart like this:

image

For my GPU and various models. How can I calculate the speed of the model in FPS? I have tried to do this that way, but I don't know that it has sense and is correct:

I run mAP calculation for my validation-set and get this information in the output: image

So if I have in my validation-set 300 of images, that model speed = 300/28 = 10,71 FPS

Is it correct or not ?


Solution

  • @Steve Zaretii

    Thanks for your answer, I ask AlexeyAB from Darknet the same question and he told me as you that my solution is limited by HDD / SDD disk.

    So he enable now solution for this type of model speed measurment: https://github.com/AlexeyAB/darknet/issues/4627

    I added -benchmark flag for detector demo, now you can use command 2652263

    ./darknet detector demo obj.data yolo.cfg yolo.weights test.mp4 -benchmark

    But for very fast models the bottleneck will be in the Video Capturing from file/camera, >or in Video Showing (you can disable showing by using -dont_show flag).

    I think that it is the best solution, you only need the newest version of Darknet (from AlexeyAB).