A comparison of different face classifiers based on Local Binary Patterns for emotional expression, gender, age and race

Computer image processing is a diversified area of information technology. Automatic age estimation, emotion, gender and race recognition based on the human face is an interesting and challenging task. In addition, this knowledge can be useful in such fields as the security industry, psychology and entertainment.The purpose is to extract certain attributes of a person from a video stream or image. This process is built up from several steps: loading the input image from a sequence, face detection, cropping the desired face, preprocessing and classification. Our goal is to find the appropriate techniques both for preprocessing and classification.In the experiments below we used face cropping and Local Binary Patterns for preprocessing and Support Vector Machines for classification. We performed the observations for each partition of each aspect using different parameter configurations. The results are collected in tables.