Facial age estimation using a hybrid of SVM and Fuzzy Logic

This paper has presented a method of facial age estimation using a hybrid of Support Vector Machines (SVMs) and Fuzzy Logic (FL). The proposed method has taken facial features from wrinkles and skin color on the human face to estimate the age group and age in point. SVMs was used to estimate the five age-groups of human age. Then, FL was implemented to estimate the age in point corresponding in each group resulting from SVMs. For performance evaluation, k-fold cross validation was carried out using FG-NET and PAL databases consisting of 700 and 500 faces, respectively. The proposed method was evaluated in comparison with five advanced methods in literature. The results showed that the proposed method provided 88.84% and 90.88% of accuracy in aging group estimation in FG-NET and PAL databases, respectively. In addition, the proposed method reported 4.81 and 3.12 for MAE (mean absolute error) for point age estimation using FG-NET and PAL, respectively. In this regard, the proposed method provided the higher performance on accuracy and MAE superior to the compared methods.