Autori
Mercaldo, Francesco
Santone, Antonella
Martinelli, Fabio

Titolo
A method for eye gaze detection through deep learning
Periodico
Rivista di Digital Politics
Anno: 2025 - Fascicolo: 2/3 - Pagina iniziale: 489 - Pagina finale: 502

In educational settings, student engagement is a critical fac- tor influencing learning outcomes, with engaged students often demon- strating better academic performance. However, real-time and large-scale tracking of student engagement poses significant challenges, especially in virtual classroom environments. Traditional methods such as direct ob- servation or post-lesson surveys are subjective, resource-intensive, and not scalable. To address these limitations, this paper proposes a novel approach leveraging the You Only Look Once model, a state-of-the-art object detection model, to monitor student eyes and assess engagement levels in real-time. This approach is based on the correlation between eye region direction and cognitive engagement, where students focusing on lesson materials or the instructor are considered engaged. The proposed model captures continuous video streams, detects students’ eye regions, and classifies their engagement into three categories: engaged, nominally engaged, and disengaged. The proposed method offers a monitoring stu- dent attention approach, allowing educators to adjust their strategies and improve the learning experience in both physical and virtual classrooms.



SICI: 2785-0072 (2025)2/3<489:AMFEGD>2.0.ZU;2-B
Testo completo: https://www.rivisteweb.it/download/article/10.53227/119822
Testo completo alternativo: https://www.rivisteweb.it/doi/10.53227/119822

Esportazione dati in Refworks (solo per utenti abilitati)

Biblioteche ACNP che possiedono il periodico