I'm currently doing real time object detection with the help of pi camera using pre-defined weights of darknet and coco dataset using openCV. Currently I'm getting 0.14 fps and my video is too much laggy. is there anyway that I could increase my fps >=20 or to some value where video is not too much laggy.

I'm using raspberry pi 4 model B with 8Gb Ram.

1 Answer 1


YoloV3 uses Darknet 53 as its backbone which is pretty heavy for a Raspberry. Maybe consider using another backbone like TinyYoloV3

  • i need accuracy in my object detection that's why i chosed yolov3 and not yolov3 tiny
    – D3mon
    Commented Dec 2, 2021 at 16:28
  • Well its a trade-off between accuracy and detection speed. Maybe lower the resolution of the video if you increased the input size of your Yolo algorithm.
    – ghostiek
    Commented Dec 2, 2021 at 16:41
  • can you tell me how to lower resolution?
    – D3mon
    Commented Dec 2, 2021 at 16:48
  • Here's some things you can do raspberrypi.stackexchange.com/questions/94698/… I've never had to lower resolution with a camera so I can't help with that sorry
    – ghostiek
    Commented Dec 3, 2021 at 12:33
  • You could get a faster computer. A high-end mini PC would cost between $500 - $900. A tradeoff of money vs speed.
    – NomadMaker
    Commented Aug 30, 2022 at 1:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.