I am currently working on face detection using viola and jones approach.
I think I am a making some silly assumptions which end up with classifiers of 100% FALSE negative rate and 0% true positive rate.
Here is how i generated my training features: for all the four feature type, i generated different sizes but of equal width and height for both black and white region.
Could this be the reason behind the poor training result?
Moreso, my threshold for each feature was arbitrarily chosen to be within a range of value.
Am I getting some things wrong? I would be so glad if someone could help. I am wasting so much time over this and exam is nearby.
