Student Engagement And Disengagement Image Dataset For Educational Research

Abstract

This study addresses the challenge of detecting student engagement and disengagement in online learning, a critical issue in the current era of virtual education. To support the development of machine learning models that can accurately identify student behavior during online sessions, we created the Student Engagement and Disengagement Image Dataset, comprising 16,000 images of students aged 5-21+ years. Images were systematically pre-processed, resized to 256 × 256 pixels, and organized into folders based on age and gender. Fine-tuning these models with our dataset improved their performance in classifying engagement and disengagement behaviors. The dataset’s structured design and extensive labeling make it an essential tool for machine learning applications aimed at enhancing student monitoring in online education. Our findings highlight the dataset's potential to contribute to more personalized and adaptive e-learning environments, though future work incorporating multimodal data could further improve model accuracy and applicability.