Hello,
Could anyone explain how the weak classifiers are constructed as decision trees and how they are formulated to calculate the error?
How is CART structure built and associated with Adaboost training logic?
Thank you very much.
Note: I asked similar questions under another title(http://answers.opencv.org/question/24147/understanding-traincascade-code/). But I think it is better to open a new subject for this. I hope it won't be a problem.