Facial emotion profiling based on emotion specific feature model

Abstract

Facial emotion profiling is rapidly becoming an area of intense interest in machine vision society for decade. In spite of major efforts, there are several open questions on how to embed the emotional intelligence in machine to respond immediately and precisely over facial expressions. In this sense, this paper presents an automatic facial emotion profiling from emotion specific feature model. A 17-point feature model on the frontal face region is proposed to track per frame fsacial emotion robustly. A measurement vector is formed based on a set of geometric distance displacements of a pair of feature points between neutral and expressive face frame. A two-stage fuzzy reasoning model is pro- posed to classify universal facial expressions. In the first stage measurements are fuzzified and mapped onto an Action Units (AUs) and later AUs are fuzzified and mapped onto an Emotion in the second-stage of fuzzy reasoning model. The overall performance of the proposed system is evaluated on two publicly available facial expression databases, average emotion recognition accuracy of 91% was achieved for RaFD and 94% for CK + database.