Innovatrics upgrades active biometric liveness detection with randomness, better UX

Innovatrics has introduced a new version of its biometric liveness detection technology to ease regulatory compliance and avoid user drop-offs. The global market for face biometric liveness checks is on track to more than double from 2025 to 2027, according to the 2025 Face Liveness Market Report and Buyer’s Guide from Biometric Update and Goode Intelligence, so the pressure to compete for market share is on liveness providers now.

The company explained and demonstrated the new active liveness in a recent online workshop.

Viktor Bielko, Innovatrics product manager for IDV solutions explained that while Innovatrics offers “semi-passive” smile and passive liveness detection methods, the company plans to replace its Eye-Gaze and MagnifEye liveness technologies with the new “Multi-Range liveness.”

Eye-Gaze liveness asks the user to follow a dot on the screen, and measures their response to the challenge. This method was not effective for all users in the wild, according to Bielko. MagniEye was launched in 2023 based on Innovatrics’ face and iris biometrics.

Both Eye-Gaze and MagnifEye are deprecated, with support lasting until the end of 2026. Multi-Range Liveness is available for integration by developers and businesses now.

Market feedback made clear that Innovatrics customers were in many cases being asked to use liveness detection with an element of randomness as a regulatory requirement. Innovatrics also found that the completion rate for its active liveness methods was not as high as the company wanted, leading it to improve the user experience.

The new liveness method also simplifies the decision for relying parties by removing a choice between two technologies.

The inclusion of randomness prevents the use of pre-recorded videos and provides defense against injection attacks, Bielko says. While dedicated injection attack detection (IAD) checks that the signal comes from the genuine device camera, and deepfake detection works by analyzing the content and sometimes its metadata, the emphasis of regulators on randomness in presentation attack detection (PAD) indicates the frequency with which these sophisticated fraud attacks are occurring.

Innovatrics tested six different actions with volunteers representing a full range of demographic groups, and found that the only one all volunteers were able to complete is taking selfies at different distances. The capture process takes longer, 15 seconds on average, but Innovatrics decided the trade-off of a few extra seconds for universal usability is well worthwhile.

Multi-Range Liveness can be configured to capture anywhere from four to eight photos, depending on the degree of randomness required. Innovatrics applies passive liveness detection to the photos, evaluates whether the challenge sequence was followed with behavioral analytics, and then returns a liveness score if the criteria is met.

Integration is essentially the same as Innovatrics’ other liveness software.

The launch of the upgraded liveness software comes amid Innovatrics’ rebranding of the DOT (Digital Onboarding Toolkit) as the IDV Toolkit.

Single image liveness

The commonality of sophisticated biometric spoofing attempts means that further research and development of PAD software to protect against 2D attacks has to provide value by offering different capabilities, like running in constricted environments where other models, like Innovatrics’, can’t run.

To that end, a pair of researchers with Rakuten have developed a model for “Robust Face Liveness Detection for Biometric Authentication using (a) Single Image” that they say uses on a single end-to-end CNN for near real-time liveness detection running on a CPU.

The lightweight model proved accurate against a new 2D spoof attack dataset created by the researchers, with processing completed in two seconds or less. The new dataset consists of more than 500 videos of 60 subjects.

The model is effective against wrap attacks as well, the researchers say, in which an image is wrapped around the head of a mannequin.

Article Topics

biometric liveness detection  |  biometrics  |  digital identity  |  face biometrics  |  Innovatrics  |  onboarding  |  user experience