Research Of Facial Expression Recognition Algorithm Based On Convolutional Neural Network

Face contains abundant information,such as gender,age,facial expression,etc.In interpersonal interaction.the facial expression plays an important role.Along with the development of artificial intelligence.facial expression recognition is widely used in many senarios,representative products such as assessment,video analysis,safe driving,emotional robot,etc.At present,facial expression recognition technology is developing rapidly and has great potential,it is a reseach hotspot in the field of human-computer interaction.Let the machine try to understand the human emotions and observe the unique human intelligence,become more humanistic so as to take place of human to complete tasks and better serve human,is always the goal that people have being pursuing,its application value is undoubtedly very important in the era of intelligence.In recent years,the excellent performance of convolutional neural network in the field of image recognition has made it a current research hotspot.Compared with the traditional recognition method that requires manual definition of features to extract features,convolutional neural network has great advantages in these aspects.This paper studies the current popular convolutional neural network algorithm and applies it to facial expression recognition.The main work of the paper is as follows:(1)This paper expounds the development status of facial expression recognition techlogy at domestic and abroad in recent years,introduces the origin and development process of neural network and the achievements made,and analyzes in detail the structure and algorithm principle of several classical convolutional neural network models.(2)This paper introduces the emoji database widely used in the industry detailly,enhances the selected emoji data set,expands the emoji database,and carries out a series of ode processing.(3)The VGGNet and ResNet network models are improved,the alley plot core structure is introduced,the MCNN network model is designed as well.Comparative experiments are conducted,and the influence of network depth,network structure,image data and other factors on the recognition effect is discussed.(4)According to the designed network model,the facial expression recognition system is established,which can automatically recognize facial expressions and display the classification results,demonstrating the network recognition effect.