Hello all,
I'm trying to run OpenCV on a 640 x 480 video feed. Of course this runs a bit slow (~10 FPS on my macbook) compared to lower resolutions, but it allows me to detect small faces, or faces far away. What I'm wondering is if I can increase performance by running it on a faster CPU, or is the bottleneck somewhere else? I did a test in OpenFrameworks as well, but the fps was similar, so I guess it's the performance of OpenCV that I need to improve on somehow. Is there some part of the OpenCV process flow I could skip, perhaps? Or might it be running faster on a pc?
Any help would be greatly appreciated!

Hi

If I make a very simple OpenCV example(e.g. read, blur, display), I get 7-8 fps with the most crappy $15 USB logitech quickcam, I get 27+ fps using my built in iSight.
This is all 640x480 and running on a MacBook Air 1.8GHz 2 GB mem(can't get smaller/slower than that before it is a netbook)

I did some hand recognition stuff and as stuff evolved I used opencv.ROI(rect) to cut down the area that opencv processes and this gave me a higher fps.

Try posting some code from where you think there is a bottleneck.