*2.1. Related Work on Vision based Clinical Screening*

Vision-based clinical screening using RGB and thermal image sensors have recently attracted increasing attention in academia and industry. Ming-Zher Poh et al. developed a robust method for measuring HR and HRV from digital RGB video recording of skin color changes [15]. He Liu et al. proposed a novel method using dual cameras to estimate arterial oxygen saturation [16]. Philips Research has been launching an app called "*Vital Signs Camera*" in 2012. Moreover, the thermal camera-based approaches have been widely applied in clinical screening and research, such as fever screening and human pose estimation [5]. To enable such specific applications, image processing method for keypoint detection has been proposed using a stacked hourglass network and feature boosting networks [17–19].

#### *2.2. Overview of Infectious Screening System using RGB-thermal Image Sensors*

In our previous work, a dual image sensor-based infectious screening system was developed for predicting the possibility of infection [10]. It comprises an RGB camera and an IRT for measuring HR, RR and body temperature. We used DFK23U618 (The Imaging Source Co. Ltd., Germany) as the RGB camera and FLIR A315 (FLIR Systems, Inc., USA) as the IRT. The visible video was recorded at a speed of 15 frames per second (fps) with a pixel resolution of 640 × 480 and the thermal video was recorded at a speed of 15 fps with a pixel resolution of 320 × 240. An RGB camera senses fluctuations in hemoglobin absorption derived from the volumetric change in facial blood vessels and obtains heartbeat signals. An IRT detects temperature changes between inhalation and exhalation in the nasal or oral area. In addition, the facial skin temperature is measured by the IRT. Multiple vital signs distinguish between patients with influenza and healthy subjects. Figure 2 shows an overview of an infectious screening system.

**Figure 2.** Overview of measurement principle that remotely senses multiple vital signs and an example of screening result.
