Lobe Software's Accuracy in Analyzing Human Facial Expressions

Abstract

The face constitutes a research subject for analyzing human facial expressions, as it can provide insights into an individual's emotions. Facial expressions are identified through changes in key facial features such as the eyes, eyebrows, mouth, and forehead. The field of education has witnessed rapid technological advancements, especially in light of the pandemic. Consequently, the need for the development of technology has become increasingly pressing. This study aims to evaluate the accuracy of Lobe software in analyzing human facial expressions. The sample population for this research comprised students from IKIP Muhammadiyah Maumere, Indonesia, with a sample size of 19 selected using a simple random sampling method. The results of the analysis showed that the accuracy of the software was pretty good, with a score of 90% for each category of neutral, sad, happy, and angry faces. The questionnaire data analysis yielded a score of 84%, which is only 6% lower than the score achieved through the students' self-reporting, implying an error rate of less than 10% if validated.