How Circadify Detects Deepfakes
Physiological liveness verification powered by real blood flow analysis — capabilities that synthetic media cannot defeat
Core Detection Capabilities
Every capability anchored in one unforgeable biological signal: real blood flow
Hemodynamic Liveness Signal
Circadify extracts the blood volume pulse waveform from facial video using remote photoplethysmography. Each heartbeat pushes blood through capillaries near the skin surface, causing micro-color changes that cameras can capture. Deepfakes, masks, and screen replays produce no such waveform — making this a binary alive-or-not signal.
Temporal Pulse Consistency Analysis
Real blood flow follows the rhythmic patterns of a beating heart — variable, organic, and physiologically constrained. Circadify analyzes the temporal consistency of pulse signals across facial regions, detecting inconsistencies that reveal synthetic generation, video splicing, or face-swap artifacts.
Multi-Region Vascular Mapping
Blood flow is not uniform across the face. Circadify maps pulse signals across forehead, cheeks, and periorbital regions simultaneously, verifying that spatial blood flow patterns match what real human vasculature produces. Spoofing tools that inject a synthetic pulse into one region fail this multi-point consistency check.
Injection Attack Detection
Virtual cameras and video injection tools allow attackers to stream deepfake content directly into verification systems. Circadify detects these attacks by analyzing whether the video stream contains genuine physiological signals or digitally constructed approximations — catching injections regardless of visual quality.
Core Detection Capabilities
The Science Behind rPPG Fraud Detection
Understanding why blood flow is the definitive anti-deepfake signal
Remote Photoplethysmography for Security
Remote photoplethysmography (rPPG) detects blood volume changes through subtle skin color variations captured by standard cameras. Originally developed for contactless health monitoring, Circadify applies this science to identity security — using the presence or absence of genuine blood flow as an unforgeable liveness indicator.
Signal Isolation and Noise Rejection
Real-world video contains ambient lighting changes, compression artifacts, and user movement. Circadify's algorithms isolate the blood volume pulse signal from all of this environmental noise, extracting a clean physiological waveform that can be definitively evaluated for authenticity — even in challenging capture conditions.
Edge-Deployable Architecture
Liveness detection runs locally on the capture device with no video transmitted to external servers. This preserves user privacy, eliminates network latency, and prevents man-in-the-middle attacks on the verification stream. Biometric video never leaves the device where it was captured.
The Science Behind rPPG Fraud Detection
rPPG Liveness vs Traditional Anti-Spoofing
| Feature | TryFaceScan | Traditional Methods |
|---|---|---|
| Detects AI-Generated Deepfakes | ||
| Passive (No User Action Required) | ||
| Works with Standard RGB Cameras | ||
| On-Device Processing | Varies | |
| Catches Video Injection Attacks | Limited |
Experience TryFaceScan Today
See what contactless vitals monitoring can do — try it free.