Research questionHow reliably can contactless thermal video estimate cardiorespiratory and sudomotor signals for operator-state awareness?Thermal cameras capture facial temperature changes without visible illumination, but those changes mix respiratory, cardiac, and sweating-related signals. Region selection, latency, polarity changes, subject variability, and limited frame rate complicate dependable physiological estimation. Latest papersRecent research connected to this question, newest first.Thermal Imaging for Contactless Cardiorespiratory and Sudomotor Response MonitoringThe study evaluates 288 region-of-interest and signal-processing configurations across 31 sessions from the public SIMULATOR STUDY 1 driver-monitoring dataset, using contact references and lag-tolerant metrics. It reports 3.1 ± 1.1 bpm breathing-rate error, 13.8 ± 7.5 bpm heart-rate error, and best fixed electrodermal-activity correlation of 0.40 ± 0.23, with individual sessions reaching 0.89. Findings are limited by the 7.5 Hz thermal-camera frame rate, region selection, latency, polarity changes, and subject variability.research paper · Sep 2, 2026