Oz Biometry

Biometric face recognition with the highest accuracy: 99.99% in NIST FRVT.

моноблок

Biometric
identification

Developed by Oz Forensics specialists, the biometric module incorporates the latest practices in artificial intelligence and is consistently improved by continuous data enrichment. Face recognition algorithm from Oz Forensics achieved one of the best test results in accuracy on LFW dataset in University of Massachussets tests in 2020 MIT LFW. The Oz Biometry module allows to identify people with a speed of less than 1 second and with 99.87% accuracy

Since the beginning of the 2010s, every smartphone has a front camera, which makes facial biometrics a natural way for authentication. Face biometry can be used both in the registration process, and in the process of searching and authentication on large biometric databases.

Facial biometrics are used in the KYC process to compare selfie photos with photographs from documents, as well as to confirm the presence of a user in biometric “black” and “white” lists.

Oz Forensics Face Recogntion Algorithms allow:

  • Detect and highlight the best shot from the video;
  • Compare face from photo with face photo from ID;
  • Perform a search in the biometric database.

Use in KYC processes

Selfie - Photo from document or from database photo

Comparison of a client's selfie with a photo from a document or from a photo stored in a government database in the process of client identification.

Selfie - Photo previously uploaded to the server

Comparison of the client's selfie with his previously uploaded photo to the server during authentication.

Selfie - Photo base "Blacklist"

Search for matches of the client's selfies in the “Black List” face database to identify fraudsters.

Selfie - Photo base "White list"

Search for matches of the client's selfie using the “White List” face database, for example, to determine the vip person.

Industry test scores (NIST, MIT):

What different datasets are tested on?

VISA

Pairs of ISO type images - ISO/IEC 19794-5 Full Frontal. Pose is generally excellent. The image type is applicable to portrait for passport, driver license, and “mugshot” images. This dataset excludes "mugshot" images.

Only adults’ images.

Mugshot

Pairs of ISO type images - ISO/IEC 19794-5 Full Frontal. Pose is generally excellent. The image type is applicable to portrait for passport, driver license, and “mugshot” images. This dataset includes only "mugshot" images.

Only USA adults` citizens images.

Visaborder

Pairs of the following images:

the first image - the image is taken from web-camera. This image is "partially" frontal and is done under the following conditions: different lighting, different angle of head rotation or partial overlap of the faces;

the second image is ISO type image - ISO/IEC 19794-5 Full Frontal . Pose is generally excellent. The image type is applicable to portrait for passport, driver license, and “mugshot” images. This dataset excludes "mugshot" images.

Only adults’ images.

Wild

Pairs of images that are very unconstrained, with wide yaw and pitch pose variation. Faces can be occluded, including hair and hands.

Only adults’ images.

99.87%

ACCURACY

99.9998%

ACCURACY

VISA

99.9903%

ACCURACY

VISABORDER

99.9936%

ACCURACY

MUGSHOT

99.9501%

ACCURACY

BORDER

99.7090%

ACCURACY

WILD

How we help you succeed>

Simple integration

Oz Forensics has open APIs for automated banking systems

Visual reporting

All reports are provided in both graphic and text form

High precision

99,99% rate of positive results

Flexibility

You can use one or all modules as needed

Delivery options

We support access via the cloud, virtual machines and software installations

Ease-of-use

Our interface is easy to use and easy to learn

All products include:

Simple,
modularized APIs

Flexible SDKs for iOS,
Android and web

Easy integration
using Docker

Get a demo >