AI-Powered Liveness Detection Prevents Identity Fraud
Innovatrics’ industry-leading facial liveness detection is based on deep neural networks to detect and prevent remote identity verification frauds. Verified by iBeta, it is compliant with ISO 30107-3 Level 2 Presentation Attack Detection (PAD) testing.
With a proven track record of biometric projects delivered globally to various government and enterprise customers, Innovatrics’ digital onboarding and liveness detection technologies support various use cases such as e-KYC, customer onboarding, bank account opening, SIM card registration, employee registration, driver registration, e-VISA issuance, digital certificate issuance, and more.
Our Digital Onboarding Toolkit offers flexibility for integrators on how they integrate and deploy technology for face presentation attack detection.See e-KYC references
Leveraging in-house R&D, we are constantly working to improve our algorithms to fight fraudulent attempts. Comprising several neural networks, our liveness detection is trained to detect various types of presentation attack vectors such as print attacks, dolls, 3D silicone masks, and display attacks.
Cooperating continuously with our clients throughout the project life cycle, we can provide recommendations for implementation and configuration, as well as a swift response to quickly changing attacks for our algorithms to meet their current needs.
Our liveness detection can be fine-tuned to meet the requirements of a specific use case – be its priority reinforced security, seamless user experience, or chosen type of deployment.
Eliminating friction for users, Innovatrics liveness detection works with industry-leading speed and accuracy, resulting in higher completion rates of new customer onboardings.
When developing the components of Digital Onboarding Toolkit, we always have user experience in mind and aim to remove as much friction from the onboarding process as possible.
Built upon deep neural networks, our passive liveness check can distinguish a real face from an image with top-performing accuracy without any user interaction.
With only one frame needed to tell if the person is real and alive, the check is instantaneous and extremely difficult to spoof.
Our semi-passive liveness detection combines the security of active liveness detection where a user is required to perform some kind of action (respond to a challenge) with little or no user-experience trade-off.
Our active liveness check requires users to perform the simple task of following a moving dot on a screen. Our algorithm monitors the movement of the pupils, evaluating whether the dot has been followed correctly.
We are constantly working on improving our liveness algorithms to keep them capable of detecting regularly evolving spoof attacks and preventing fraud while making sure they are not ethnically, racially, or sexually biased.
Much attention is given to building and using our own datasets when training neural networks for presentation attack detection. Moreover, we have established thorough liveness testing carried out at our own R&D center to deliver the best performing and unbiased algorithms possible.Learn more in our white paper
The proprietary technology developed by industry leader
Works on-device or on server-side
Works on web, Android, iOS