Combining Heritance AdaBoost and Random Forests for face detection

AdaBoost has proved to be an effective method to improve the performance of base classifiers both theoretically and empirically. However, previous studies have shown that AdaBoost might suffer from the overfitting problem, especially for noisy data. In addition, it still needs much time to train the classifier using AdaBoost. In this paper, we focus on designing an algorithm named Heritance AdaBoostRF to solve the two problems. We use Heritance AdaBoost to enhance the detection speed instead of AdaBoost, and Random Forests as the weak learners of Heritance AdaBoost to deal with the overfitting problem. Experiments clearly show the superiority of the proposed method over MIT-CBCL face database and a highest detection rate of 98.94% is obtained, and experimental results based on unbalanced data sets of MIT+CMU face database showed that the overfitting problem has been improved effectively.