Liveness detection is a set of techniques used during biometric capture to verify that the presented sample originates from a live, present human rather than a spoof such as a photograph, mask, recording or deepfake. It distinguishes genuine presentations from presentation attacks by analysing physiological signals, motion, texture and challenge responses, and is standardised under ISO/IEC 30107 as presentation attack detection. Liveness detection is essential to the integrity of remote identity verification, biometric authentication and onboarding flows.
Overview
- Liveness detection addresses the core weakness of biometrics: a captured face or fingerprint can be presented by an attacker.
- Approaches split into passive methods, which require no user action, and active methods, which issue a challenge.
- ISO/IEC 30107 formalises presentation attack detection and its evaluation metrics.
- Modern systems combine multiple cues and increasingly rely on Deep Learning classifiers.
Mechanisms
- Texture and reflectance analysis: distinguishing genuine skin from printed or screen-rendered surfaces.
- Motion and depth cues: detecting micro-movements, 3D structure and parallax that flat spoofs lack.
- Challenge-response: prompting blinks, head turns or random expressions to defeat static attacks.
- Anti-spoof classification: Convolutional Neural Network models trained on attack and genuine samples.
Applications
- Remote Know Your Customer onboarding for financial services.
- Face unlock and Biometric Authentication on mobile devices.
- Border control and high-assurance Access Control.
- Fraud Detection in account recovery and high-value transactions.