Research Summary: A vision-based behavioural monitoring framework towards trustworthy AI proctoring for online assessments

This research outlines a vision-based framework for precise head pose estimation, integrating face detection, facial landmark detection, and 3D pose estimation. The methodology demonstrates high accuracy across its components, making it suitable for applications requiring robust human behavioral monitoring, such as human-computer interaction, driver monitoring, and security surveillance.

Why it matters

Accurate and robust head pose estimation is critical for advancing technologies that rely on understanding human attention and behavior. This framework provides a foundational capability for developing more sophisticated and reliable monitoring systems across diverse sectors, enhancing safety, interaction quality, and surveillance efficacy.

Key insights

• A face detection model achieved 94% accuracy with high precision, recall, and F1-score.
• A pre-trained Tensorflow CNN model for facial landmark detection yielded an F1-score of 0.89 and a mean localization error of 1.5 pixels.
• Mathematical techniques were used for 3D pose estimation, resulting in low mean absolute errors for pitch and yaw (2.1 degrees).
• The integrated framework provides accurate head pose estimation for various monitoring applications.

Source

Frontiers in Education — https://www.frontiersin.org/articles/10.3389/feduc.2026.1900028