L2CS-Net
L2CS-Net predicts the gaze direction of a student. Attentiveness will be assumed if the student is looking forward.
AI-Powered Attention Monitoring System for Enhancing Online Learning Engagement
AI-Driven Program to Boost Student Attention in Online Classes
Tam Cheung Yan Shojin
Supervisor Prof. Zhao Hengshuang
This project proposes an innovative application that leverages computer vision AI models, including L2CS-Net for gaze detection, STAR Loss for head pose and facial landmarkdetection, ResEmoteNet for emotional analysis, and YOLOv11 for real-time phone detection.The application aims to monitor student attentiveness via camera input, enhancing online learning experiences by providing real-time feedback on engagement levels.
Models
Star Loss
Star Loss detects head poseand facial landmarks. Attentiveness is calculated based on drowsiness and head direction.
Yolov11
Yolov11 is employed for real-time phone
detection.
ResEmoteNet
ResEmoteNet predicts the emotions of the
student. The model categorises emotions as one of the seven classes: angry, disgusted, fearful, happy, neutral, sad, and surprised.