Look-Up Table Tabanli Boosting Algoritmasi ile Karmaşik Sahnelerde Yüz Sezimi Face Detection in Cluttered Scenes Using Look-Up Table Based Boosting Algorithm
Serdar Kargın, Bülent Sankur, Hava Harp Okulu · 2005
In this work, a classifier which can detect human faces on still gray images is proposed. Constrcution of the classifier is based on using Look-Up Table (LUT) based Gentle AdaBoost algorithm. Each classifier stage is a binary tree classifier which each node is a strong classifier constructed using GAB algorithm and computationally efficient simple Haar like features. Each node and tree stage classifier use the output of the previous classifier of its own type. Another important feature of the classifier construction algorithm can handle the variations in training set by using automatic division scheme and can be trained using large number of images. Thereby image classes which shows large in-class variations can be classified with low false alarm rates.