AI changing biometrics

Source: Deep Learning on Medium


Biometrics has become an integral part of people’s lives due to development of new technologies. What can be called Biometrics 2.0 have also more usage of such technologies as AI, creating new possibilities and value.

In general, biometrics is a system for recognizing people based on one or several physical (or behavioral) characteristics. Typically, the approach of operation of biometric systems is reduced to two main types.

The first is called verification, a comparison test result with a biometric template. This option helps to check if this is the person for whom he/she claims to be.

The second one is identification. After receiving a specific sample, the system is verified with the biometric database to determine the identity.

There are some huge projects based on biometric technology. Perhaps one of the most ambitious is a project implemented by AADHAAR in India.It is a biometric identification system that contains data on more than a billion people.

According to experts of the Acuity Research during the 2018 in the world will be about 3.5 billion electronic documents.

Biometrics 2.0

As the technological development continues, the shift towards contactless methods of biometric recognition can be noticed. In Biometrics 2.0 the contactless methods dominate.

The most notable AI technologies in Biometrics 2.0. are image recognition, machine learning and deep learning and they also contribute to development of new biometric techniques such as vein recognition, facial expressions, ear shape and even chip implantation.

In Russia, biometric technologies are evolving quite rapidly. For example, the largest Russian banks are working with the biometric identification systems, such as customer identification system by voice and pictures.

Face recognition as a service

The Russian AI-company RecFaces could be called the Biometrics 2.0 company. The company uses deep learning in their core business.

RecFaces developed an information platform for multimodal identification, called Id-Me.

The system removes data from cameras and instruments using a capture module (ID-box). It converts the image and sound into a specialized index. RecFaces works simultaneously with several biometric metrics.

On the server, the biometric data is mapped to a database of user profiles or unwanted persons. Recognition is provided by deep learning algorithms.

RecFaces Id-Me biometric platform is designed to suit banks, airports, retail, hotel business, sports organizations, and government agencies.

Banks can use biometrics to increase security. A possible case of an attempt by a fraudster to withdraw money from someone else’s card. An ATM camera connected to Id-Me identifies the face of the person trying to do this. If this information does not coincide with that contained in the database, withdrawals are almost instantly blocked. To use this method of protection, it is not even necessary to equip an ATM with additional equipment.