Tech5 claims victory in contactless fingerprint liveness competition

Tech5 has announced its mobile fingerprint liveness detection won first and second place in the systems category of the Liveness Detection (LivDet) 2023 Non-contact Fingerprint international competition.

The neural network-based liveness detection system submitted by Tech5 placed first and second for both single and four-fingerprint algorithms, according to the announcement, with the lowest error rates for bona fide biometrics. The company says its technology, with a model only 1.5MB in size, detected complex spoofs with Ecoflex, Playdoh, wood glue, and latex with 100 percent accuracy, demonstrating both accuracy and efficiency.

LivDet 2023 was organized by Clarkson University and the University at Buffalo. The competition tested presentation attack detection (PAD) technologies on a variety of known and unknown presentation attack instruments and live fingers.

“We are delighted to share these fantastic results of our research team’s effort, leading to a development of algorithms that are twenty times more accurate than the nearest competition in the evaluation, making it effectively the best NN-based fingerprint liveness detection technology in the world,” says Rahul Parthe, co-founder, chairman, and CTO of Tech5.

Tech5’s contactless fingerprint technology is marketed as the T5-AirSnap Finger software for mobile devices. The tech has been deployed to the market, and is also being trialled for biometric self-enrollment by the UK Home Office. The company has also worked with BixeLab to evaluate the accuracy of contactless fingerprints and their compatibility with contact-based reference templates.

LivDet has not yet posted the official results of the competition, but will announce the results at the IJCB 2023 conference in September.

Article Topics

accuracy  |  algorithms  |  biometric liveness detection  |  biometric testing  |  fingerprint biometrics  |  presentation attack detection  |  TECH5