Difference in time for YOLOv3

asked 2018-09-28 06:12:18 -0600

alex_s gravatar image

updated 2018-09-28 06:20:42 -0600

Hi, I run dnn/object_detection.cpp with params: -c=yolov3.cfg -m=yolov3.weights -classes=coco.names --scale=0.00392 --width=416 --height=416 and it works, but time it takes for one image is about 4 seconds. Function: net.forward(outs, getOutputsNames(net)) takes the most time. Although on the image (https://drive.google.com/open?id=1kZI...) on top is written about 745 ms. I run in release and use opencv4.

Also in this link (https://github.com/opencv/opencv/pull...) developers writes that median time per 416x416 image is 216.11ms.

What may be the reason of this time difference?

I have Ubuntu OS: https://drive.google.com/open?id=1Mvb...

edit retag flag offensive close merge delete

Comments

i removed your screenshots, they are useless here.

(can't be indexed, quoted, and simply eat folk's bandwidth)

please edit your question, and add information as text if nessecary, thank you.

berak gravatar imageberak ( 2018-09-28 06:15:08 -0600 )edit

please, again:

you should want to make it as easy as possible to read your post, and also in response quote from it.

your screenshots are defeating that purpose, and putting them on an external bin makes it even worse.

berak gravatar imageberak ( 2018-09-28 06:23:59 -0600 )edit