Research Of Deep Learning For Facial Expression Recognition

Facial expression recognition is a very important and popular research topic in the academic field,especially in the field of artificial intelligence and machine learning.If the computer is expected to have real intelligence,then the computer must first learn to understand the human feelings,and the facial expression often contains rich emotional information.It can be seen that facial expression recognition is an integral part of the computer's future intelligence development.At present,many researchers have been involved.Compared to the traditional facial expression recognition methods which need artificial definition and artificial feature extraction,a great advantage of deep learning is to remove the shortcomings which require a large number of human factors intervention,thus reducing the uncertainty of expression recognition.In recent years,deep learning has achieved excellent results in many fields,such as computer vision,Natural Language Processing and so on.Therefore,this paper studies the popular recognition algorithm based on depth learning,and applies the face verification algorithm based on deep learning to apply it to facial expression recognition.The main work of this paper is reflected in the following aspects.(1)a facial expression recognition algorithm based on deep learning is proposed.By using the YOLOv2 algorithm to improve the features of facial expression recognition,the algorithm can detect the position of the face and identify the facial expression type by input of the picture or video,and the improved algorithm has a better recognition.At the same time,the rate of recognition will not reduce the speed of recognition.(2)a face verification algorithm based on depth learning is proposed,which improves the face verification of the most accurate Deep ID2 algorithm on the video again,and combines the long and short period neural network to achieve the identity fusion of different timing,thus improving the accuracy of face verification in the re video.Based on that,parallelization is applied to further improve the operation speed of the algorithm.(3)an expression recognition prototype system based on deep learning is realized.Based on the main workflow of facial expression recognition system,the overall architecture of expression recognition system is designed,and an expression recognition prototype system based on depth learning is realized.The proposed facial expression recognition algorithm based on depth learning improves the accuracy of facial expression recognition,increases the average accuracy by 2%,and runs faster than the current popular expression recognition algorithm.And through the combination of face verification algorithms based on depth learning,different expressions of the same person can be obtained,which can provide help for further analysis of facial expression.