Detecting Student Engagement in Online Learning Using AI-Based Body Language Analysis

In online learning environments, the lack of physical interaction limits educators’ ability to assess student engagement through non-verbal cues. This paper presents a deep learning approach for automatic engagement detection using body language analysis. A hybrid CNN-LSTM model is developed to extract spatial features from video frames and capture temporal behavioral patterns. The system is trained and evaluated on the DAiSEE dataset, achieving an accuracy of 97.0%, significantly outperforming baseline models. Evaluation metrics, including precision, recall, and F1-score, confirm the model’s effectiveness across engagement categories. The proposed method offers a robust solution for understanding student engagement in virtual classrooms. Future work will explore real-time deployment and multimodal fusion to enhance adaptability in live learning platforms and support personalized, engagement-aware education. Index Terms—Online Learning, Student Engagement, Deep