Facial expression recognition Using Machine Learning

Abstract

Facial expression recognition (FER) plays an important role in computer vision. In this paper, we compare the result of two proposed methods of representing Facial landmarks detection and Feature Extraction. The first method is based on image processing (for example, histogram equalization, thresholding, color conversion, morphological operations, etc.) and the second one used the Dlib library to detect facial landmarks. We examine each feature descriptor by considering two classifications methods such as Support Vector Machine (SVM) and the Multi-layer Perceptron (MLP) with three facial expression databases(10k US Adult Faces Database, the MUG Facial Expression and personal database) to classified three different facial expressions: happiness, surprise and neutrality.The Experimental results demonstrate that the first proposed method shows 91.5% accuracy and more than 96% accuracy for the second method.