Location

Hilton Hawaiian Village, Honolulu, Hawaii

Event Website

https://hicss.hawaii.edu/

Start Date

3-1-2024 12:00 AM

End Date

6-1-2024 12:00 AM

Description

Instructors often use facial cues of their students as key indicators of student attention levels. However, this method can pose a problem in online and computer-based learning environments. While other research has shown computer vision and eye-tracking could be used with machine learning techniques to predict attentiveness, they have shown only moderate success in terms of accuracy. In this work, we improve upon existing techniques for student attention tracking. We employed our previously developed Non-Intrusive Classroom Attention Tracking System (NiCATS) to collect facial images and eye-tracking data of students during three controlled experiments that represent common academic scenarios. Our first contribution is using convolutional neural networks to predict student attentiveness with an F1-Score of 0.91. Our second contribution is the validation of using eye-tracking metrics in conjunction with machine learning models to predict the attentiveness of students with up to 0.78 F1-Score, which could be useful when webcam privacy is a concern.

Share

COinS
 
Jan 3rd, 12:00 AM Jan 6th, 12:00 AM

Utilizing Convolutional Neural Networks and Eye-Tracking Data for Classroom Attention Tracking

Hilton Hawaiian Village, Honolulu, Hawaii

Instructors often use facial cues of their students as key indicators of student attention levels. However, this method can pose a problem in online and computer-based learning environments. While other research has shown computer vision and eye-tracking could be used with machine learning techniques to predict attentiveness, they have shown only moderate success in terms of accuracy. In this work, we improve upon existing techniques for student attention tracking. We employed our previously developed Non-Intrusive Classroom Attention Tracking System (NiCATS) to collect facial images and eye-tracking data of students during three controlled experiments that represent common academic scenarios. Our first contribution is using convolutional neural networks to predict student attentiveness with an F1-Score of 0.91. Our second contribution is the validation of using eye-tracking metrics in conjunction with machine learning models to predict the attentiveness of students with up to 0.78 F1-Score, which could be useful when webcam privacy is a concern.

https://aisel.aisnet.org/hicss-57/ks/edtech/2