image processing - OpenCV HOG and SVM implementation -
i trying use hog , svm methods find person walks under camera. working opencv train_hog example. changed size( 96, 160 )
size( 240, 240 )
in lines 433 437 438 , 442 because size of video (240x240).
i use inria dataset negative samples , frames walking person video positive samples.
after running train_hog.cpp, got no error. problem when give video input test_it()
function, couldn't find same person walking around.
what problem here?
i have 100 positive sample. few? or problem default hog , svm parameters? (because didn't change parameters) or else?
example pos sample:
p.s : can done background subtraction want use hog , svm.
Comments
Post a Comment