I want to classify road signs using Hog descriptors and SVMs (a linear kernel). I am going to classify each sign versus a non sign (eg different training for each sign). After some research I have seen that it is better to use SVMlight instead of function SVM_train of opencv and that the hog descriptor for people detection implemented in opencv was also trained using SVMlight. Is this true? And why is it better to use SVMlight or LIBsvm instead of class SVM? Thank you!
Asked
Active
Viewed 727 times
