Detekce očí pomocí příznakového rozpoznání

Abstract

The purpose of this thesis is to test success of Haar, LBP and HOG methods using features for object detection in computer vision, especially for detection of human eyes in faces. Get to know library OpenCV dealing with computer vision, use it's implementations for creating cascade classifiers for each detection method. Experimentally test function of created classifiers on test dataset and define it's speed and accuracy. In last chapter there is comparison with classifier of library.