A simple multi-feature classification to recognize human emotions in Images

Abstract

The recognition of emotions via facial expressions is a complex process of piecing together various aspects of each facial feature. Since viewing a single facial feature in isolation may result in an inaccurate recognition of emotion, this paper attempts training neural networks to first identify specific facial features in isolation, and then use the general pattern of expressions on the face to identify the overall emotion. The technique presented is very basic, and can definitely be improved with more advanced techniques that incorporate time<br>and context.