A Real-Time Multimodal Human Fatigue Monitoring System Using GSR and Vision-Based Facial Analysis

Human fatigue and drowsiness can reduce attention and task performance, motivating thedevelopment of real-time monitoring systems that combine behavioral and physiological indicators. This paperpresents a low-cost prototype that combines galvanic skin response (GSR) sensing with vision-based facialanalysis. An R-1085 GSR sensor is interfaced to an ESP32 for analog signal acquisition and USB serial transmission.A conventional laptop camera provides facial images, while Python, OpenCV and MediaPipe facial landmarks areused to estimate eye aspect ratio (EAR) and identify prolonged eye closure. LabVIEW 2022 is proposed as thesupervisory monitoring layer for visualization and alert presentation, with a buzzer available for warning output.Initial GSR experiments using 35 finger-contact samples produced values from 2242 to 2270 (mean 2262.4;standard deviation 4.75),whereas 37 no-contact samples ranged from 2366 to 2490 (mean 2400.38; standarddeviation 23.81). The camera prototype produced EAR values of 0.39 for an open-eye frame and 0.19 for a closedeye frame. These preliminary results demonstrate operation of the individual sensing components. The prototypeis intended for research and educational evaluation; the measured thresholds are not clinically validated fatiguecriteria.