Non-contact physiological signal-based emotion recognition and early warning system for college students

Mental health monitoring among college students has become increasingly important, yet traditional assessment methods often lack objectivity and real-time capability. This study presents a novel non-contact emotion recognition system that integrates millimeter-wave radar, smart cameras, and infrared thermal imaging to collect physiological signals. By analyzing respiratory patterns, cardiac activity, eye movements, and facial temperature distribution through deep learning algorithms, our system achieves real-time emotion recognition and early warning. Experimental results with 120 college students demonstrate an overall accuracy of 87.5% in emotion classification, with effective early warning capabilities for abnormal emotional states. The proposed system offers a promising solution for automated mental health monitoring in educational settings.