A Case Study of Facial Emotion Classification Using Affdex

Abstract

This paper focuses on the analysis of reactions captured by the face analysis system. The experiment was conducted on a sample of 50 university students. Each student was shown 100 random images and the student´s reaction to every image was recorded. The recorded reactions were subsequently compared to the reaction of the image that was expected. The results of the experiment have shown several imperfections of the face analysis system. The system has difficulties classifying expressions and cannot detect and identify inner emotions that a person may experience when shown the image. Face analysis systems can only detect emotions that are expressed externally on a face by physiological changes in certain parts of the face.

Artificial Intelligence Techniques for Distance Education: A Systematic Literature Review

“…Many of the systems depend on the databases created by the researchers to achieve their systems and train it, as Ngoc Anh collected his data by taking snapshots of students in lessons (the subject of basic programming) at FPT University [30]. Other researchers also recorded pictures of students in different poses with many facial expressions and collected them in the databases of their systems [31], [33], [35].…”

“…1) The student's sitting position and the angles of the body are difficult to explain whether the body is tilted or otherwise to judge if the student watches the video online or not [4], [30]. 2) Lack of categorized reactions and feelings, as one of the expressions that do not find any similar classification which may appear [33], [35]. 3) Facial expressions vary very quickly and with many different feelings [33].…”

Generating Robotic Speech Prosody for Human Robot Interaction: A Preliminary Study

“…This toolkit is one of the most used toolkits that is a cross-platform to recognize multi-face expression in real-time by using the facial action coding system (FACS). It has been showing a high accuracy and reliable system for several application [38][39][40]. The length of time for each recognized emotions was calculated and analyzed to determine if the emotions presented in the robot's voice reflected on the emotional states of users.…”

AutisMitr: Emotion Recognition Assistive Tool for Autistic Children

“…An individual can accurately predict emotions by observing facial expressions and somatic changes by transforming these noted changes to their physiological presentation. According to Darwin's study, it is presumed that the process of identifying emotions involves multiple models of behaviour, thus providing a detailed classification of 40 states of emotions [33]. However, the majority of studies related to stratification of facial attributes refers to the classification of six basic emotions presented by Ekman [34].…”