An Intelligent Decision Framework For Facial Expression Recognition Based On Deep Learning

Abstract

Abstract Facial expression recognition is very important for computer interacting with human. In this paper, the intelligent decision framework based on deep learning is proposed to improve the accuracy of facial expression recognition, which employs three channels to extract different complementary deep features. Each channel has the combination of different convolution layers and MaxPool layers and alleviating the gradient disappearance problem by residual network to obtain the possible expression. Finally, the results of three channels are employed to obtain the final facial expression by the intelligent decision function. Experiments on the public datasets of Fer2013 and Afew5.0 show the effective results.