A synchronized multimodal dataset for unobtrusive cardiorespiratory monitoring

Non-contact and continuous vital-sign monitoring systems offer the potential for more accessible and inclusive measurement. However, publicly available datasets to support their development remain limited, particularly those that provide synchronized radar and Red-Green-Blue (RGB) camera recordings alongside electrocardiogram (ECG) derived heart rate and respiration-belt-derived respiratory rate reference signals. In this study, we present a synchronized multimodal dataset for vital-sign monitoring, acquired using both radar and RGB camera sensors and complemented by reference vital-sign measurements. The dataset includes recordings collected under diverse breathing conditions\u2014resting, post-exercise, and guided patterns spanning 4 to 30 breaths per minute. Each recording contains datastreams from three investigative sensors, including raw radar data, remote photoplethysmography (PPG) extracted from the RGB camera, and wrist-based PPG and inertia measurement unit (IMU) measurements. In parallel, several reference datastreams were collected, including the respiration waveform, pulse rate from a pulse oximeter, and the ECG signal. In total, 451 recordings were collected from 31 participants across all experimental conditions.