DEFORGIMAL - DEtection of FORGed Image credentials using MAchine Learning

iProov is a pioneer in the field of identity verification, delivering an outstandingly simple and
easy user authentication experience, by means of face verification on mobile devices and
laptops. It is protected against the chief threat to face verification - hacker spoofing attacks -
by a unique method. The DEFORGIMAL project is an innovation in the area of Adaptive
Biometrics, which will make the iProov system much more robust against sophisticated
attacks. The iProov service is provided as a cloud-based identity verification service to
security-conscious service providers, such as financial institutions, online education providers,
subscription service providers and web services requiring precise identification of users, as
well as to secure employee access in enterprises large and small. Password–related login
mechanisms have long been problematic in these contexts, and a biometric login service has
distinct advantages, provided it can be made resiliently secure. The DEFORGIMAL project
will research and combine machine learning algorithms into a closed loop feedback system to
provide robust and adaptable detection of novel attempts to spoof the iProov face verification
system and continuous adaptation to such threats. This satisfies the requirements of long term
users, who need confidence in the ability of the iProov system to adapt to an environment of
changing threats. Today, although substantial academic research has been carried out on
protecting face verification against spoofs, and on adaptive biometrics for an individual’s
data, there are no solutions available that provide adaptation of the sort of generic spoof
detection filters iProov has pioneered. DEFORGIMAL is a proof of concept project, which
will enable the enhancement of iProov’s product to address the full range of authentication
needs in the market and provide confidence to security-savvy prospective customers
in advance of commercial implementation.