Modern educational institutions face the challenge of understanding and responding to the emotional states of students in real time. Traditional schooling approaches rely heavily on teacher observation, which is inherently subjective, scalable only to small groups, and limited to explicit behavioural cues. This research proposes an Emotion-Aware School System (EASS) that integrates Artificial Intelligence (AI) with Internet of Things (IoT) devices to continuously monitor, analyse, and respond to student emotional states across classroom and campus environments. The proposed system employs a Convolutional Neural Network (CNN)-based facial expression recognition model deployed on Raspberry Pi edge nodes. These nodes stream anonymised emotional classification outputs to a centralised Spring Boot backend, which aggregates real-time emotion data and surfaces actionable intelligence to teachers, counsellors, and administrators through a Flutter-based dashboard. The system recognises seven core emotional states — Happy, Sad, Angry, Fearful, Disgusted, Surprised, and Neutral — mapped to pedagogical intervention categories. Privacy is preserved through on-device inference; raw facial images are never transmitted over the network.
Tambe et al. (Mon,) studied this question.