Human Age Estimation using Facial Landmarks

Human age can be inferred by distinct patterns emerging from facial appearance. Usually people wear makeup aimed at making them look very much beautiful, handsome and younger than real appearance making hard to predict the age. The photo of a person with makeup and without makeup can lead us into different answers because makeup affects the looks of the person by covering dark and red spots, wrinkle, discoloration area, breakouts and other undesirable areas on the face. In this paper, high quality images with and without makeup are collected from internet ranging from 35 to 55 years in age. Selecting 20 most important facial feature points manually using statically developed Active Appearance face model. The input image is subjected to pre-processing to make it suitable for further image processing.1D-Gabor filter are used for feature extraction. We use Local Gabor XOR pattern (LGXP) operator to encode Gabor phase and efficient in extraction of texture from facial images. Feed forward neural network for age estimation.