An Efficient Support Vector Machine Algorithm for Age and Gender Detection from Face Pictures
Thanh Tran · 2021
In this paper, a novel kernel function is proposed to perform an efficient Support Vector Machine, SVM, classification for age and gender estimation and detection. A robust, easy-to-execute algorithm is proposed based on the SVM as a core method to simultaneously predict the age and gender of highly variable pictures. Owing to its efficiency because of kernel function proposed in this paper, the SVM based algorithm is able to powerfully classify the features, leading to more accurate and precise estimation of age and gender. The algorithm is modified to solve the simple problems as well as complex situations, estimating age and gender simultaneously. The mean error of proposed method is between 5.2% – 7.3% that the less accurate results higher ages are due to lack of sample pictures in training phase. To be fair, all of the comparisons with other methods were done on the same database, FG-NET, including 1002 pictures with different qualities, diverse styles, and various rate of face coverage. The simulation results show that the proposed method has lower mean error in comparison with other recent related works.