DOI: 10.4018/IJSSCI.2019010101
International Journal of Software Science and Computational Intelligence
Volume 11 • Issue 1 • January-March 2019
Copyright © 2019, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
1
Unobtrusive Academic Emotion
Recognition Based on Facial
Expression Using RGB-D Camera
Using Adaptive-Network-Based
Fuzzy Inference System (ANFIS)
James Purnama, Swiss German University, Indonesia
Riri Fitri Sari, University of Indonesia, Depok, Indonesia
ABSTRACT
Quality of learning in the classroom is influenced by many factors. One of them is the academic
emotions of the students. The emotion detection in the classroom cannot be done by using sensors
attached to the body of the students, because it would disturb the concentration of the students. The
proposed solution is by using unobtrusive emotion detection, e.g. by placing video capture equipment,
which is not visible at the front of the student’s desk. In this study, an RGB - Depth Microsoft Kinect
camera is used to record facial expressions by considering the convenience factor of the students,
speed of response time, and cost efficiency. A combination of Cohn-Kanade dataset and EURECOM
dataset is used as the training set in machine learning with Adaptive-Network-Based Fuzzy Inference
System (ANFIS) algorithm, with 8 sample of Asian race students (4 male and 4 female students).
KEywoRDS
Academic Emotion Detection, ANFIS, Facial Expression, RGB-D Camera
INTRoDUCTIoN
In the era of Information Technology, the teaching and learning process in the classroom is still the
main method of learning. In addition, there are e-learning methods and Intelligent Tutoring System,
which are supplements to the classroom learning method. Quality of learning in the classroom is
influenced by many aspects, one of which is the Academic Emotions.
Pekrun et al. (2002) describes that the Academic Emotions can affect learning qualities which
confidence, excitement, frustration, interest, flow/engagement, boredom, confusion, and anxiety
(Arroyo et al., 2009; Azcarraga et al., 2011; Burleson, 2006; Azcarraga et al., 2010; Kapoor et al.,
2007; Zeidner, 2007).
Previously, emotion detection used sensor devices that are plugged / affixed to the body of
the research subject. The sensor devices are physiological signals or biosignal sensors to measure
body signals, such as Electromyography (EMG), Electro Dermal Activity (EDA) or Galvanic Skin
Response (GSR). This method is commonly referred to as the Hidden Ways (covertly). This method
was perceived only suitable for testing in the laboratory, and are not effective when implemented in