Open Access

ARTICLE

VMFD: Virtual Meetings Fatigue Detector Using Eye Polygon Area and Dlib Shape Indicator

Hafsa Sidaq¹, Lei Wang¹, Sghaier Guizani^2,*, Hussain Haider³, Ateeq Ur Rehman^4,*, Habib Hamam^5,6,7

1 Key Laboratory for Ubiquitous Network and Service Software, School of Software, Dalian University of Technology, Dalian, 116024, China
2 College of Engineering, Alfaisal University, Riyadh, 11533, Saudi Arabia
3 Institute of Human-Centred Computing (HCC), Graz University of Technology, Inffeldgasse 16c, Graz, 8010, Austria
4 School of Computing, Gachon University, Seongnam-si, 13120, Republic of Korea
5 Faculty of Engineering, Uni de Moncton, Moncton, NB E1A3E9, Canada
6 School of Electrical Engineering, University of Johannesburg, Johannesburg, 2006, South Africa
7 Research Unit, International Institute of Technology and Management (IITG), Av. Grandes Ecoles, Libreville, BP 1989, Gabon

* Corresponding Authors: Sghaier Guizani. Email: ; Ateeq Ur Rehman. Email:

Computers, Materials & Continua 2026, 86(3), 25 https://doi.org/10.32604/cmc.2025.071254

Received 03 August 2025; Accepted 06 October 2025; Issue published 12 January 2026

Abstract

Numerous sectors, such as education, the IT sector, and corporate organizations, transitioned to virtual meetings after the COVID-19 crisis. Organizations now seek to assess participants’ fatigue levels in online meetings to remain competitive. Instructors cannot effectively monitor every individual in a virtual environment, which raises significant concerns about participant fatigue. Our proposed system monitors fatigue, identifying attentive and drowsy individuals throughout the online session. We leverage Dlib’s pre-trained facial landmark detector and focus on the eye landmarks only, offering a more detailed analysis for predicting eye opening and closing of the eyes, rather than focusing on the entire face. We introduce an Eye Polygon Area (EPA) formula, which computes eye activity from Dlib eye landmarks by measuring the polygonal area of the eye opening. Unlike the Eye Aspect Ratio (EAR), which relies on a single distance ratio, EPA adapts to different eye shapes (round, narrow, or wide), providing a more reliable measure for fatigue detection. The VMFD system issues a warning if a participant remains in a fatigued condition for 36 consecutive frames. The proposed technology is tested under multiple scenarios, including low- to high-lighting conditions (50–1400 lux) and both with and without glasses. This study builds an OpenCV application in Python, evaluated using the iBUG 300-W dataset, achieving 97.5% accuracy in detecting active participants. We compare VMFD with conventional methods relying on the EAR and show that the EPA technique performs significantly better.

---

Keywords

---