EMOTION RECOGINATION USING LDA AND CNN IN E-LEARNING

Abstract

Facial expression is one of the most powerful channels of non-verbal communication, contains PLENTY of information that can help in determining the mood/emotion of a person. Recognition of facial emotions can be used for a variety of applications. Our face is a gold mine for emotion markers, and by recognizing these distinct markers for each emotion we can efficiently determine the emotions of people. Our main aim in this paper is to extract the important features of the face which can help in detecting the emotion of the person in question (e.g. Pupil size, mouth, nose etc.). We first use, LDA (Linear Discriminant Analysis) algorithm for extracting the face of an individual from the picture/video. It acknowledgment errands and hence improves performance. LDA algorithm is used due to its higher discriminatory power. Once we extract the face from the picture/video we need to extract the necessary features using the CNN (Convolutional Neural Network) algorithm, after which we classify them into different emotion groups based on the extracted features again using CNN algorithm. Highlight extraction and choice in example acknowledgment have been an essential issue and have been talked about much of the time. Additionally, two- dimensional flags, for example, difficult to be demonstrated well by conventional models like SVM. Based on our understanding, we have explained these concepts in the paper.