Low Resolution Facial Expression Recognition Based On Teaching Environment

As a base technology of face recognition,facial expression recognition is indispensable in the research field of automated facial analysis.Market development and practical applications based on facial expression recognition technology are developing briskly,and in fields as interaction of human-computer,model of social network,medical treatment,communications and automatic drive,this technology has been turned into successful applications or services.Meanwhile,more application scenarios are being explored,and the application in education is not fully exploited at this stage.For computers,the task of facial expression recognition is usually to classify images or video frames into several categories such as anger,disgust,fear,happiness,sadness and surprise.In this paper two facial expression recognition methods are proposed.The first is facial action unit assisted method,which combines convolutional neural network with Bayesian network.The other one is residue learning method based on generative adversarial network.Researches and validations are made upon facial expression databases and these methods perform well.Moreover,the researches combine realistic scene,apply it into student facial expression recognition in primary school classroom in order to analyze the status of students in low resolution,concentrating on exploring low resolution facial expression images.