Here are my hints on how to speed up the Face Detection. First try to use the Local Binary Patterns Cascades coming with OpenCV. As you can see in this post, calculating the LBP codes for an image is computationally very simple. You can find the available LBP Cascades in the data
folder coming with your OpenCV installation, see:
If this doesn't speed up the detection process significantly or if you are using LBP cascades already, then try to downscale the image with cv::resize
, but I can't make any assumptions on the minimal image size for your detection task. Anyway, downscaling the input images should speed up the detection definitely.
And the most obvious way to speed up the detection process is to probably use a highly parallelized CascadeClassifier
as available in the GPU module: