NEC : develops face presentation attack detection technology that conforms to ISO standards in iBeta testing

Tokyo, October 11, 2021 - NEC Corporation (NEC; TSE: 6701) announced today the development of a highly accurate face presentation attack detection technology against photos, displays, 3D masks, etc. when performing face recognition. This technology has been tested by iBeta (*1), an international independent third-party testing and quality assurance organization, and conforms to standards established by the International Organization for Standardization (ISO). Moreover, the technology is optimized for use on mobile terminals, and contributes to improving the reliability of online identity verification and other face recognition applications that are expanding in use.

iBeta offers a variety of software testing and quality assurance services globally, and is accredited by the authoritative U.S. National Institute of Standards and Technology (NIST, *2), especially for biometric quality assurance. The testing of face presentation attack detection is based on ISO standards with an attack presentation classification error rate (APCER) of 0% for face photographs and face images on displays (Level 1), and an APCER of 1% or less for 3D masks and mannequins (Level 2).

NEC's highly accurate face presentation attack detection technology was developed utilizing know-how cultivated in the research and development of NEC's face recognition technology, which has the world's No. 1 authentication accuracy (*2) and is at the core of NEC's portfolio of biometric authentication technologies, "Bio-IDiom" (*3). This technology complies with Level 1 of iBeta's presentation attack detection software testing, as well as Level 2. Furthermore, the bona fide presentation classification error rate (BPCER) is less than 5%, which is far below the iBeta standard of 15% or less. In general, low APCER and BPCER are incompatible, but NEC's technology has achieved both.

In addition, the presentation attack detection process has been accelerated, and it can be operated with limited computer resources, such as smartphones and tablets. This enables smooth face presentation attack detection and prevention, even when performing face recognition on mobile terminals through payment applications and websites, thereby contributing to online identity verification with both high accuracy and reliability.

In the future, in addition to applications for mobile terminals, NEC aims to use this technology for a wide variety of purposes, including access gates for companies, large facilities, and event venues, as well as for payment terminals and specialized services.