00. Overview¶

Who should read this page¶

This page is for anyone who wants the fastest orientation before going deeper.

What this documentation is¶

This repository explains face liveness for eKYC in simple language while still covering the points that matter in real systems.

Face liveness answers one core question:

Is the camera seeing a real person who is physically present right now, or is it seeing some kind of spoof?

A spoof could be:

a printed photo
a replayed video on another screen
a mask
an injected image or video stream
AI-generated or manipulated face content

Why this repo was reorganized¶

A lot of face liveness material becomes hard to read because it mixes everything together too early:

definitions
attack taxonomy
metrics
deployment engineering
standards and compliance
vendor evaluation

That makes the first reading heavy.

This repo now uses a simpler structure:

Main guide¶

Use this first. It explains the core ideas in plain English.

Appendix and deep reference¶

Use this when you need technical detail, standards context, long checklists, or deeper testing material.

The big idea to remember¶

A face match system and a face liveness system solve different problems.

Check	Main question
Face match	Does this face look like the enrolled face or the ID portrait?
Face liveness	Is this a real live person present during capture?

A system can be good at face matching and still accept a spoof.

That is why remote eKYC usually needs both.

Reading path by role¶

Role	Start with	Then go to
Product or business	01-face-liveness-guide.md	02-ekyc-integration.md, 04-best-practices.md
ML engineer	01-face-liveness-guide.md	03-deployment-guide.md, appendix/A3-metrics-and-evaluation.md
Backend or platform engineer	02-ekyc-integration.md	03-deployment-guide.md, appendix/A5-security-and-privacy.md
Risk or fraud team	01-face-liveness-guide.md	appendix/A2-attack-taxonomy.md, appendix/A5-security-and-privacy.md
Compliance or procurement	01-face-liveness-guide.md	appendix/A4-standards-and-compliance.md, appendix/A6-vendor-evaluation-checklist.md

The main guide by phase¶

Phase	What it includes
Start Here	01. Face Liveness Guide, 02. eKYC Integration, 03. Deployment Guide, 04. Best Practices
Practical Playbook	05. Real-World Examples, 06. API and Response Examples, 07. Decision Logic, 08. Evaluation Playbook, 09. Common Failures, 10. Product Guide
Engineering and Operations	11. Advanced Topics, 12. Fusion and Meta-Model, 13. Dataset Strategy, 14. Score Calibration and Thresholding, 15. Error Analysis, 16. Monitoring and Operations, 17. Security Hardening, 18. Device and Platform Matrix, 19. Model Governance
Support and Architecture	20. FAQ, 21. Troubleshooting, 22. Case Studies, 23. System Architecture
Appendix	A1 to A10 for glossary, metrics, standards, data policy, and experiment design

Need term help early?¶

If terms like APCER, BPCER, calibration, virtual camera, or policy engine are unfamiliar, keep these nearby while reading advanced pages:

Visual reading map¶

flowchart TD
    A[Start] --> B[00 Overview]
    B --> C[01 Face Liveness Guide]
    C --> D[02 eKYC Integration]
    D --> E[03 Deployment Guide]
    E --> F[04 Best Practices]
    F --> H[05 to 11 Practical Playbook]
    H --> I[12 to 19 Engineering and Operations]
    I --> J[20 to 23 FAQ, Troubleshooting, Cases, Architecture]
    C --> G[Appendix A1-A10]
    D --> G
    E --> G
    F --> G
    H --> G
    I --> G
    J --> G

A simple mental model¶

Think about face liveness as one layer in a larger trust pipeline:

flowchart TB
    A[1. Capture]
    B[2. Quality checks]
    C[3. Liveness]
    D[4. Face match]
    E[5. Risk policy]
    F[6. Decision]

    A --> B --> C --> D --> E --> F

Face liveness is important, but it is not the whole system.