Evaluation Method of Teaching Styles Based on Multi-modal Fusion

Abstract

Teaching style refers to the teaching performance of the teacher's personal characteristics, teaching skills and teaching methods formed in the long-term teaching process. It plays an important role in helping students achieve academic success. With the continuous reform of the teaching system, more and more courses are taught in the form of videos. In this paper, we established a teaching style evaluation system and classification model based on teachers’ teaching behavior based on facial expressions, voices and postures in the teaching process. We adopted principal component analysis and autoencoder feature selection methods, and designed based on The multi-modal step-by-step fusion algorithm of deep neural network classifies teaching styles.