Engagement Estimation for Intelligent Tutoring System in E-learning Environment

Sheharyar, Imranul Haq, Jia-cai ZHANG

Abstract


Task engagement is the effortful concentration used to strive towards task goals. Engagement of student during lecture is very important for grasping and understanding the concepts and it significantly improves student performance. Engagement detection thus automatically offers opportunities for better learning, especially in online learning such as MOOCs learning. One of the most important aspects that is missing from the online learning platforms is real time feedback depending on the state of the students. In our daily life facial expressions are source of conveying emotions and are considered to be the primary means of communication. In this study we have proposed a system that can capture the facial cues of the student and estimate the engagement of the student during online learning. The accuracy we have got for binary classification is 84% with 0.72 kappa and for multiclass we got 60% accuracy with 0.53 kappa.

Keywords


Histogram of Oriented Gradient (HOG), Gabor, Local binary points (LBP), Scale invariant feature transform (SIFT) and Principal component analysis (PCA)


DOI
10.12783/dtssehs/iceme2019/29563

Full Text:

PDF