EmoAI Smart Classroom: The Development of a Student Emotional and Behavioral Engagement Recognition System
Abstract: Emotions significantly influence the learning environment, impacting student engagement and the overall educational process. With the rise of challenges in maintaining student engagement in large offline classrooms due to various factors, there's an increasing need to detect classroom emotions. This study aims to detect and analyze emotions evoked in classrooms to enhance the educational experience for both educators and learners. Utilizing the DAiSEE dataset, which captures various affective states in offline classroom settings, an engagement detection system was developed using the YOLOV8 state-of-the-art model and deployed on Roboflow.
The system's framework was based on capturing facial cues and was augmented with an interactive interface for lecturers.
Despite its advanced capabilities, the model achieved a precision of 74.5%, a recall of 60.6%, and a mean average precision (mAP) of 65.3%. The findings suggest that while the model offers significant insights, there's potential for further refinement, particularly given the limited frames used for training. The study's interactive interface offers real-time feedback for lecturers, underscoring the intertwined relationship between emotions and learning. Future directions include real-time engagement detection and alert systems, emphasizing the potential to revolutionize classroom dynamics through emotionally attuned educational environments.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Yonatan_Bisk1
Submission Number: 1971
Loading