Adaptive System for Real-Time Attentiveness Detection in Online Meeting Platforms

The concept of online classes and meetings received a significant push during the pandemic, offering people greater convenience at work and enabling students to attend sessions from their comfort zones. However, this shift also brought shortcomings, including reduced attentiveness and easy misuse of attendance systems. To address these issues, this paper proposes an adaptive system that detects the attentiveness of meeting attendees in real time. The system triggers an alarm at the attendee’s end upon detecting inattentiveness, while also providing the organiser with attendance data reflecting each participant’s attentiveness score. The detection is carried out using the Dlib algorithm, which first performs face detection followed by facial feature extraction for the eyes and mouth. Eye movement analysis and yawning detection are then used to classify the attendee’s state as Active, Drowsy, or Sleepy. The Eye Aspect Ratio and Mouth Aspect Ratio, computed through Euclidean distance measures on facial landmarks, serve as the primary indicators. Experimental results show that the combined eye and yawning detection achieves an accuracy of 95.32%, demonstrating the system’s suitability for enhancing online meeting engagement and attendance management.