简介:Thisstudyanalyzeslivefacialvideosforrecognizingnonverballearning-relatedfacialmovementsandheadposestodiscoverthelearningstatusofstudents.First,coloranddepthfacialvideoscapturedbyaKinectareanalyzedforfacetrackingusingathree-dimensional(3D)activeappearancemodel(AAM).Second,thefacialfeaturevectorsequencesareusedtotrainhiddenMarkovmodels(HMMs)torecognizesevenlearning-relatedfacialmovements(smile,blink,frown,shake,nod,yawn,andtalk).Thefinalstageinvolvestheanalysisofthefacialmovementvectorsequencetoevaluatethreestatusscores(understanding,interaction,andconsciousness),eachrepresentsthelearningstatusofastudentandishelpfultobothteachersandstudentsforimprovingteachingandlearning.Fiveteachingactivitiesdemonstratethattheproposedlearningstatusanalysissystempromotestheinterpersonalcommunicationbetweenteachersandstudents.