New approaches for gender, emotion and age prediction from facial images

There are many studies that are done and made on face images by image processing. Nowadays, studies on tissue analysis are increasing day by day and are progressing rapidly to provide a popular solution. In this study, four different new methods based on computer vision have been proposed as an alternative to the classical diagnostic methods for age, gender and emotion estimation. Emotion, age and gender classification process used face images of the FEI database. Four different new approaches based on Local Binary Pattern (LBP) have been proposed to extract effective features from color facial images. Recommended methods are used for color images. In order to obtain the best classification from the data, data were moved to different color spaces and different noise methods were applied to the data. The determination of the success rates in the classification process was carried out in two stages. In the first stage, the attributes were obtained and in the second stage, these attributes were graded by 10-fold cross validation test using different machine learning methods. At the end of the study, 99% of the gender classification, 99.5% in the age classification and 97.5% in the emotion classification were achieved.