Biometric Sensor Network Tracks Student Engagement in Real-Time

A new thesis by Ahmed Elsayed proposes a Biometric Sensor Network (BSN) designed to measure individual student engagement in STEM classrooms in real-time. The system relies on distributed Student Processing Units (SPUs) that act as sensing nodes, capturing behavioral, emotional, and cognitive indicators through cameras. Unlike traditional methods, the network prioritizes five core objectives: non-intrusiveness, non-invasiveness, non-stigmatization, real-time operation, and automation—while maintaining rigorous data security and privacy.

Each SPU operates in two modes. In dataset-collection mode, raw student video is temporarily recorded to build a private engagement dataset for training models. In analysis mode, the SPU performs on-device inference on 10-second video segments, running face detection, gaze estimation, and affective analysis locally. This design ensures no identifiable video frames ever leave the device. A secure backend handles authentication, session orchestration, and encrypted data ingestion. The integrated system spans hardware, computer vision, wireless networking, and security protocols, offering a scalable approach to automatic, privacy-preserving engagement measurement in live lecture environments.