A Deep Learning-Based Model for Student Engagement Detection in E-Learning Environments to Enhance Cognitive Skills
Download| Volume 6 Issue 1, 2025 | |
|---|---|
| Author(s): |
Ali Aijaz Shar Shah Abdul Latif University Khairpur Mir’s, aliaijazsharbalouch@gmail.com Samina Rajper Shah Abdul Latif University Khairpur Mir’s, samina.rajper@salu.edu.pk Noor Ahmed Shaikh Shah Abdul Latif University Khairpur Mir’s, noor.shaikh@salu.edu.pk |
| Abstract | This research introduces an advanced deep learning framework for real-time monitoring of student engagement in digital learning environments. The system employs a hybrid convolutional neural network (CNN) and support vector machine (SVM) pipeline trained on a purpose-built dataset comprising 100,000 annotated video frames from 50 undergraduate participants. Experimental results demonstrate exceptional performance, achieving 92.5% accuracy in facial emotion recognition and 87.3% precision in binary engagement classification ("engaged" vs. "disengaged"). A critical innovation involves gender-aware personalization modules that attained 95% identification accuracy, enabling tailored pedagogical approaches that elevated cognitive skills by 13.4% relative to control groups. The architecture processes video streams at 15 frames per second using standard hardware resources through an optimized Python-based interface leveraging TensorFlow and OpenCV libraries. This means that achieving this efficiency allows for seamless integration into existing in-house learning management systems. During validation, adaptive interventions based on engagement metrics—content simplification, motivational feedback, and interactive quizzes showed a substantial improvement in critical thinking (+14.2%) and problem-solving (+13.1%) competencies. Without any physiological sensors used, the solution overcomes the basic limitations of conventional e-learning systems by conducting continuous, non-intrusive assessment. The study further reveals behavioral insights: male participants exhibited 23% more neutral expressions during complex tasks, while female students' greater emotional variability enhanced model sensitivity. These findings validate computer vision analytics as a scalable mechanism for personalized education, with immediate applications in virtual classrooms and professional training platforms. Future work will explore cross-cultural validation and multimodal sensor integration to enhance generalization. |
| Keywords | Student engagement detection, convolutional neural networks, adaptive e-learning, real-time facial analysis, emotion recognition, educational technology |
| Year | 2025 |
| Volume | 6 |
| Issue | 1 |
| Type | Research paper, manuscript, article |
| Recognized by | Higher Education Commission of Pakistan, HEC | Category | Journal Name | ILMA Journal of Technology & Software Management | Publisher Name | ILMA University | Jel Classification | -- | DOI | - | ISSN no (E, Electronic) | 2790-590X | ISSN no (P, Print) | 2709-2240 | Country | Pakistan | City | Karachi | Institution Type | University | Journal Type | Open Access | Manuscript Processing | Blind Peer Reviewed | Format | Paper Link | https://ijtsm.ilmauniversity.edu.pk/arc/Vol6/i1/pdf3.pdf | Page | 20-29 |