A gender classification method using age information

Facial features used for gender classification are affected by their aging process, because human´s face is gradually changed as they grow up. Thus, in this paper, we propose a gender classification method robust to age variation by using age information and two facial features: appearance and geometry feature. Local Binary Patterns (LBP) is used as an appearance feature to classify gender of young and adult age group, and Euclidean distance among facial feature points is used as a geometry feature to classify gender of old age group. Experimental results showed that performance of our proposed method is increased about 2% compared to gender classification without using age information.