CONIX Publication

Pedagogical research, especially regarding university instructors, indicates that classroom nonverbal behaviors are critical for encouraging and maintaining student engagement. Movement, gesturing, and eye contact, for example, may help establish instructor credibility and positively impact student learning ability. However, it is difficult for instructors to observe and engage with data on their nonverbal behavior during lectures, currently requiring expert monitoring or extensive knowledge of classroom immediacy literature. In this research, we develop an accessible, intelligent interface to help instructors view and engage with data on their teaching nonverbal behavior. Using EduSense, a comprehensive monitoring tool that detects relevant features of classroom behavior in real-time, we develop a front-end platform that 1) Introduces key classroom behavior metrics to instructors, 2) Visualizes data outputs from the EduSense pipeline, and 3) Produces intelligent recommendations to help improve instructor nonverbal behavior. We also provide a portal for researchers to view platform engagement metrics and customize interface settings to support user testing. This research furthers development for the EduSense distributed system, building upon computer vision and machine learning techniques for broad scale classroom sensing.