parallelization of apps/traincascade/cascadeclassifier::fillPassedSamples()

edit

Hi,

I am wondering if anyone has successfully parallelized the outer most loop of apps/traincascade/cascadeclassifier.cpp::fillPassedSamples()

This becomes particularly important when training a cascade with a large number of negative samples and a deep cascade. I've tried to follow the inner workings of the various class methods being invoked such as imgReader, featureEvaluator. Its not clear to me whether or not wrapping the outer loop in a parallel_for can work.

The following link shows how to address the innermost loop however that doesn't make a significant impact. http://answers.opencv.org/question/16...

Any help is much appreciated. -Paul