Principle teaches Boosting algorithm, the basic concept of AdaBoost algorithm, training algorithm, compared with random forests, training error analysis to derive generalized additive model, exponential loss function, training algorithm, selection, sample weights of weak classifiers weight reduction, practical application .
AdaBoost algorithm it is the most typical application is the visual target detection, such as face detection, pedestrian detection, vehicle detection, and so on. Prior to deep learning popular, with these simple features plus AdaBoost classifier to do target detection, it is always a mainstream program our industry, in academia made it inside the paper the most.
Outline:
Labs
Application Brief
VJ framework Introduction
cascade classifier
Haar features
principle training algorithm
to train their own model
a variety of VJ framework of improved
overall summary
Experimental aspects: