ABSTRACT
A difficulty for teachers in COVID-era online teaching settings is assessing engagement and student attention. This has made adapting teaching to the responses of the class a challenge. We developed a system called Engage AI for assessing engagement during live lectures. Engage AI uses video-based machine learning models to detect drowsiness and emotions like happiness and neutrality, and aggregates them in a dashboard that instructors can view as they speak. This provides real-time feedback to instructors, allowing them to adjust their teaching to keep students engaged. There is no video data transmitted outside of students' web browsers, and individual students are anonymous to the instructor. Testing in undergraduate engineering lectures resulted in 78.2% reporting feeling at least potentially more engaged during the lecture and at least 34.4% of students reporting feeling more engaged during the lecture. These approaches could be applicable to many forms of remote and in-person education. © American Society for Engineering Education, 2021