Deep Learning-Based Facial-Expression-Aware Interface Adaptation for Virtual Assessments

This repository provides replication materials for the study “Deep Learning-Based Facial-Expression-Aware Interface Adaptation for Virtual Assessments.” The study integrated a facial-expression classifier with a Moodle plugin that adapted visual elements and motivational messages during mathematics assessments with children aged 7 to 8.The deposited materials include the facial-expression model code and anonymized results from three experiments comparing the standard and adaptive Moodle interfaces. The experiments assessed quiz performance, completion-time efficiency, perceived ease of use, and perceived usefulness.No facial images, videos, names, or other directly identifying information about the participating children are included. Facial frames used during assessment were processed temporarily for inference and were not retained. The external facial-expression dataset used to train and test the classifier is not redistributed in this repository.