Expression Analysis using kernel Extreme Learning Machines
Expression analysis is a topic covered under affective computing which is under a lot of research inthe field of computer vision. We propose an expression analysis algorithm that utilizes kernel ELMsand CNNs to determine the state of the expression. The expressions include sadness, happiness,fear, anger, disgust, surprise and neutral. The first step is to detect the face in the image and to dothat we use the DPM face detector and for extracting the features we use the VGG face network.Once we have the features of the face selected in the image we use kernel extreme learningmachines (ELM) due to it’s high speed of execution and the accuracy. Seven ELMs are needed toobtain predictors for seven expressions. The prediction is then performed using a fusion networkthat obtains features from two independent networks. Along with the input from the two networks,scores are taken from all the ELM models as input, to enhance accuracy