Term of Award
Master of Science, Computer Science (M.S.C.S.)
Document Type and Release Option
Thesis (restricted to Georgia Southern)
Copyright Statement / License for Reuse
Digital Commons@Georgia Southern License
Department of Computer Science
Committee Member 1
Committee Member 2
Gursimran Singh Walia
Nowadays, students attend school in various ways: in-person, online, hybrid. Hence, identifying student attentiveness is crucial for educators as attention matters to student success, and it is easier to teach attentive students. However, facial emotions have an influence on student attentiveness, and this attentiveness is linked to learning processes. This thesis finds a correlation between facial emotions and attentiveness, which can support automated predictive labeling of student attentiveness. Cloud-based emotion recognition platforms processed a labeled dataset of students' images, and the returned confidence values from those APIs were used to perform statistical analysis. We used a labeled dataset to examine the viability of classifying the attentiveness via machine learning models. An accuracy of 83% was achieved using XGBoost compared to other machine learning models for AWS. Additionally, regression analysis indicated that Happy/Surprise for AWS and Neutral/Happiness emotions for Azure are statistically significant indicators of attentiveness. Selecting images based on these emotion values improved the model to 90% accuracy.
Sultana, Atia, "Finding Correlation in Facial Emotions to Support Predictive Automated Labeling of Student Attentiveness" (2021). Electronic Theses and Dissertations. 2255.
Research Data and Supplementary Material