Expression Recognition using Neural Network

Abstract

A facial expression is a gesture created by the facial muscles that can convey human emotions. Facial expression recognition system is used to recognize different emotions of a face. A facial expression is exhibited by the movement of muscles underneath the face skin. Facial expression Recognition comprises of three main phases viz. face detection using Haar classifier, feature Selection using Local Binary Pattern algorithm and Expression Classification using Support Vector Machines. The image files for training are stored in the form of pixels in CSV file and are then converted by the algorithm to form images which subsequently decreases the size of the training data set. The proposed system predicts various emotions from a single image rather than predicting only one kind of emotion as done by the existing systems. Training of the system is done using the neural network.