Engagement is one of essential components in a learning process to provide personalized intervention pedagogy. Since there is a paradigm-shift on information and communications technology (ICT) in education, fostering learner’s engagement is not only for traditional classrooms but also for distance learning settings such as online learning. In this paper, we propose a practical use of a real-time engagement assessment through facial expression by optimizing convolutional neural network. We built a face recognition and engagement model to be applied in a web-based learning application. The deep learning model is experimented on open Dataset for Affective States in E-Environments (DAiSEE) with hard labeling modification. Extracting images from every 10 seconds video were done to prepare the dataset to be fed into the neural network. Some hyper-parameter tunings are conducted to achieve high accuracy (more than 90%). The engagement states are recorded to perform an evaluation of the learner’s during any online learning activities such as reading, writing, watching video tutorial, online exams and online class. The results demonstrate the potential improvement for interactive distance learning by taking into account engagement assessment.
Shofiyati Nur Karimah, Japan Advanced Institute of Science and Technology, Japan
Shinobu Hasegawa, Japan Advanced Institute of Science and Technology, Japan
This paper is part of the ACE2020 Conference Proceedings (View)
View / Download the full paper in a new tab/window