Innovatrics enhances biometric liveness detection with image quality checks
Filtering out low quality images from assessments improves the accuracy of Innovatrics’ passive biometric liveness algorithm, as the company’s Head of R&D Marian Beszedes explained when presenting the findings of its research during the EAB’s recent workshop on face image quality.
Edge cases observed by Innovatrics all share similarly low image quality, the company says in a summary of the research published on its website. Screening them out can improve performance without compromising user experience, according to Beszedes.
In a post detailing the research and the test dataset it used, Innovatrics also notes that high-quality images with extra high sharpness can also give away spoof attempts.
Integration consultant for Innovatrics’ Digital Onboarding Toolkit (DOT) Peter Molčány says in a ‘Trust Report’ article from the company that its biometric liveness algorithms are continuing to improve, even when the change is not observable, thanks to an innovative method for improving its datasets.
Innovatrics launched its own data collection app in order to target specific groups of people, in another option for the same kind of dataset quality inspection advocated by Wise above.