c++ - OpenCV - needing to detect side/front/back upper bodies(neck + head), heads(from back and front) - training classifier properly -
c++ - OpenCV - needing to detect side/front/back upper bodies(neck + head), heads(from back and front) - training classifier properly -
this first ever opencv haar classifier training , need observe people , blurr out faces/heads/upper bodies(head+shoulders) in images , taking static images, photographic camera not ever move.
i started extracting positive images sample images taken static camera. started extracting heads, shoulders + head in different orientations(back, front, sideways), since image contains people in different positions - imagine restaurant or bar.
the problem positive images not contain distinct object collection of different objects ( head, shoulder + head, side head, side shoulder+head, head, shoulder+head). trained classifier 6 phases, 50 positive images, , around 600 negative images(taken online repository). tried classifier image, , detected random non object pieces image.
i wondering, given background of problem, if can point me in right direction of doing sort of detection , training - may separating classifier in detecting heads, or shoulders , head, etc... or okay mix different positions in 1 classifier
i have tried different pregenerated classifiers came opencv , have extremely low success rate on image since image taken higher position people, , creates angle.
c++ image opencv image-processing haar-classifier
Comments
Post a Comment