Many CvSVM vectors into one... how?

asked 2012-09-12 07:58:11 -0500

milgraj gravatar image

Hi

How can I combine n vectors that i got from training my svm into only one that I can use in detecting objects by HOGDescriptor?

edit retag flag offensive close merge delete

Comments

there are many question on how to use hoddescriptor with SVM, none has a reasonable answer tho

yes123 gravatar imageyes123 ( 2012-09-13 06:25:25 -0500 )edit

What do you mean by "n vectors from the SVM"? Training an SVM should give you a single model.

StevenBell gravatar imageStevenBell ( 2013-01-04 16:27:14 -0500 )edit

I am trying to do this myself. I am not able to get it working though. I calculated the Primal Form of the support vectors from the trained model generated from CvSVM. http://en.wikipedia.org/wiki/Support_vector_machine#Primal_form

First I multiplied all coeffesients in each support vector with the alpha belonging to the support vector. Then I summarized each coeffesient of each vector together so that I got one long vector instead of many long vectors.

jompa234 gravatar imagejompa234 ( 2013-03-11 08:46:31 -0500 )edit