Towards Contactless Human Concentration Monitoring Using mmWave Signals

Maintaining concentration in today’s complex and distracting environments is increasingly challenging, with significant impacts on productivity, learning outcomes, and safety. Traditional methods like self-reporting and observational studies are subjective and labor-intensive. Current approaches, including camera-based systems and wearable sensors, raise privacy concerns and require continuous physical interaction. To address these limitations, we propose a novel, contactless concentration monitoring system using mmWave technology. Our system leverages Commercial-Off-The-Shelf (COTS) mmWave devices to detect concentration-related activities, such as eye blinking, nodding, yawning and leg shaking. In particular, we enhance the activity detection and overcome the limited field of view (FOV) of mmWave devices through spatial decomposition based on Delay-and-Sum (DAS) beamforming technologies. Moreover, we mitigate interference in concurrent activities by exploiting the distinct frequency ranges associated with each concentration-related activity based on Short-Time Fourier Transform (STFT). A CNN model, integrated with domain adaptation techniques, ensures robust performance in diverse environments. Experiments involving 10 volunteers demonstrated an overall accuracy of 95.3% in detecting human activities. The system maintained robust performance at distances up to 150 cm and across different office environments. Our method offers a contactless and privacy-preserving alternative to current approaches, making it suitable for applications such as classroom monitoring, workplace productivity observance, and cognitive health monitoring.