Video Processing-Marker Detection slow?
Hi there! I am trying to build a program using ArUco Library which tracks markers and renders virtual objects on them. However,whenever marker detection process takes place, the frame rate drops a lot. What can I do to decrease processing time? Right now I build on Debug.Will it be better on release or not? Also, what else could I do to have real time video processing? I tried using pyrdown function, but the results are almost the same.
Any advice for video processing in real time is welcome!
for realistic timing, use a release build.
Ok I used Release mode. It is much better,however I still need something faster.Any other ideas? Is it true that by using OpenCV 3.0 video processing may be better? Right now I use 2.4.10
no, that won't make much difference.
What resolution do you work on? I am sure that the processing (detection) may be done on a lower resolution, and this will make a big difference in time (except if you have done this already)
640x480 You mean using pyrdown right? Or just by passing to the processing function a smaller image?Isn't that what pyrdown does?
I was thinking at the resize. It seems that pyrDown does something else (a little bit)
Oh then you think I should resize and add a resized image for processing huh? Ok I will try it!
I have done that for making faster the face detection, so I suppose that it will improve speed in your case, too