Tracking Emotions through Facial Expressions in Online Education Systems Based on Transient Emotion Peak

Abstract

Two types of students' performance evaluation mechanisms are called formative and summative. To have good results at the summative evaluation, the students must have frequent constructive feedback throughout a course. It is not an easy task for a teacher to give frequent feedback. As a solution for that, online tests can be introduced with feedback. A constructive feedback should consist of three features, i.e. Appreciation, Advice and Evaluation. When incorporating the appreciation in a feedback in an online education system, the emotions can be analysed to appreciate the student' effort. The most feasible way to analyze the emotions is through the facial expressions in an online system. However, there should be a systematic way to track the emotions based on each question, while the students are answering. This research proposes a time interval based on the transient emotion peak considering the time taken to read a question as the stimulus activity to cause an emotional change, to capture the emotional changes. A series of photographs were captured by using a webcam for each question and analysed the emotions by using MSemotion-API. At the end, a single photograph was selected based on this time, tracking the most different emotion out of the series. A video recording was taken simultaneously, and it was analysed by a psychiatrist and kept as the benchmark dataset. At the end, a cross validation test was done based on the output of API and the benchmark dataset and was noticed that this has given a good result.