How It Works

From Camera to Vital Signs in 30 Seconds

Our rPPG technology transforms any standard camera into a contactless vital signs monitor. Here is a high-level overview of the process.

Step by step

The Complete Process

The process begins when a user positions their face in front of any standard camera — smartphone, webcam, or tablet. The camera captures video at 30 frames per second, creating a stream of images for analysis.

  • ✓Works with any camera 720p or higher
  • ✓Requires approximately 30 seconds of footage
  • ✓Ambient lighting is preferred over direct light
  • ✓User should remain relatively still

Computer vision algorithms detect the face and identify regions of interest (ROI) — areas where blood flow changes are most visible. The forehead and cheeks are primary ROI due to their vascular density.

  • ✓Real-time face tracking and alignment
  • ✓Multiple ROI analysis for redundancy
  • ✓Handles minor head movements
  • ✓Quality checks for face visibility
Face detection and region-of-interest tracking on a scanning UI

For each frame, the algorithm extracts average color values from the ROI. These RGB (red, green, blue) signals contain the physiological information — subtle color changes caused by blood flow beneath the skin.

  • ✓Separate analysis of RGB channels
  • ✓Green channel contains strongest pulse signal
  • ✓Frame-by-frame color averaging
  • ✓Spatial averaging reduces noise
Signal processing toolbox that filters noise from the raw signal

Raw signals contain noise from lighting changes, motion, and camera artifacts. Advanced digital signal processing techniques isolate the physiological signal from these contaminants.

  • ✓Bandpass filtering for heart rate frequencies
  • ✓Motion artifact removal algorithms
  • ✓Illumination normalization
  • ✓Independent Component Analysis (ICA)
rPPG signal extracted from subtle skin color changes

Our proprietary neural network analyzes the processed signals to extract vital signs. The model recognizes patterns that correlate with specific physiological parameters across diverse populations and conditions.

  • ✓Convolutional neural network architecture
  • ✓Trained on diverse populations
  • ✓Continuous model improvement
  • ✓Real-time inference capability
Neural network architecture that infers vital signs from signals

The system outputs multiple configured readings from a single scan: heart rate, respiratory rate, heart rate variability (HRV), and optional BP or SpO2 signals. Results include confidence context.

  • ✓Heart Rate (HR) in BPM
  • ✓Respiratory Rate (RR) in breaths/min
  • ✓HRV metrics (SDNN, RMSSD)
  • ✓Optional blood pressure trends
  • ✓Optional SpO2 signals
Vital sign results returned from a single contactless scan

Ready to See It Work?

Experience contactless vital signs detection for yourself with a personalized demo.

Request Demo →Try Demo →

Request A Demo

See how contactless vitals can transform your healthcare delivery.