Introduction

Biometric Anti-Fraud (BAF) is a technology stack for intelligent processing, analysis and recognition of images with an application scenario for remote human identification.

Applies AI and machine learning to recognize faces and determine such facial attributes as face liveness, image quality assessment, etc

Designed for embedding in corporate and government remote customer services

Implements cross-platform (iOS, Android, WEB) self-registration and authorization scenarios with anti-spoofing checks of the face image and checks for matching a selfie with a photo in the document

Used for authorization or registration of users, BAF provides measures to detect forgery of persons and (optionally) documents

Registration scenario uses remote facial liveness detection and identification (1:N) algorithm. Authorization scenario is implemented by remote facial liveness detection and verification (1:1) algorithm

Face and document checks

Depending on the subsystems selected and the use cases, the following checks are available:

Basic checks:

• Comprehensive selfie liveness check
• Comprehensive image quality check
• Comprehensive check of user's environment change
• Comprehensive blacklist check
• Risk alerts

Additional (optional) checks:

• Matching of a selfie face with a document face
• Document check for forgery
• Liveness Reflection check
• Motion control video saving (BETA video stream processing service. In this version, saving the video of an attempt to login or register is available)
• Cross-matching. Comparison of face best shots from recorded videos and best shot from component