Analyzing the Emotions of Individuals with Special Education Needs During the Educational Process Using Facial Recognition Software

Abstract

Facial expressions play a significant role in interpersonal interaction. The primary way individuals convey their intentions is by incorporating facial expressions into the communication process. The ability to read facial expressions is a cornerstone of social relationships, and this skill becomes vital in situations where words are insufficient or deceptive. Today, the ability for computers to rapidly scan and analyze facial expressions is facilitating human life in numerous fields. Studies in the literature indicate that this analysis is used in many fields: for security purposes, such as taking precautions against malicious individuals or detecting fatigue for safe driving; for identifying health problems, such as pain diagnosis in newborns; and for creating personalized experiences in psychological and social domains, like generating music playlists based on mood. For individuals with special education needs, interpreting emotions from facial expressions is a more challenging skill; however, understanding their emotions is of greater importance due to their specific conditions. In this study, a recognition software was developed using the Python programming language, capable of detecting the human face and identifying emotional states from facial expressions. Using this software, 10 students with special education needs were observed over a period of 8 weeks. The research methodology relies on comparing the instantaneous emotion detection records from the software (developed using Python, OpenCV, and DeepFace libraries) with the dominant emotion records identified per session by an expert human observer. The analysis results demonstrated a high degree of overlap between the emotions detected by the software and those identified by the observer. According to data from both the software (totaling 791 emotion detections) and the observer, the most frequently observed dominant emotional states among the students during the 8-week educational period were found to be "neutral" (natural state) and "happiness". These findings suggest that the developed software can be utilized as a highly efficient and valid tool for understanding the emotional states of students in special education environments.