Explaining training method for AdaBoost algorithm

Posted by konzti8 on Stack Overflow See other posts from Stack Overflow or by konzti8
Published on 2010-05-02T21:42:05Z Indexed on 2010/05/02 21:48 UTC
Read the original article Hit count: 360

Hi, I'm trying to understand the Haar feature method used for the training step in the AdaBoost algorithm. I don't understand the math that well so I'd appreciate more of a conceptual answer (as much as possible, anyway). Basically, what does it do? How do you choose positive and negative sets for what you want to select? Can it be generalized? What I mean by that is, can you choose it to find any kind of feature that you want no matter what the background is? So, for example, if I want to find some kind of circular blob - can I do that? I've also read that it is used on small patches for the images around the possible feature - does that mean you have to manually select that image patch or can it be automated to process the entire image? Is there matlab code for the training step?

Thanks for any help...

© Stack Overflow or respective owner

Related posts about adaboost

Related posts about machine