Our facial recognition algorithm is globally top-ranked by NIST in the FRVT 1:1 leaderboards.

Latest NIST FRVT evaluation report 2024-12-20

🆔 ID Document Liveness Detection - Linux - Here

🤗 Hugging Face - Here

📚 Product & Resources - Here

🛟 Help Center - Here

💼 KYC Verification Demo - Here

🙋‍♀️ Docker Hub - Here

ID-Document-Liveness-Detection-Docker

Introduction

This repo demonstrates the server-based ID document liveness detection (anti-spoofing) capabilities for ID card, passport, and driver's license.
Our ID document liveness detection SDK(ID document anti-spoofing) suite effectively prevents common presentation attacks: Screen Replay Attacks, Printed Copy Attacks, and Portrait Substitution Attacks.

◾ID Card Recognition Product List

No.	Repository	Release Type
1	ID Card Recognition - Android	Android
2	ID Card Recognition - iOS	iOS
3	ID Card Recognition - Flutter	Flutter
4	ID Auto Capture - React	Web Front-end
5	ID Card Recognition - Windows	Server-Windows
6	ID Card Recognition - Linux	Server-Linux
7	ID Card Recognition - C#	Server-Windows
➡️	ID Card Liveness Detection - Linux	Server-Linux

Docker Run

• Pull docker image.

sudo docker pull kbyai/idl:latest

• Run docker container with license key file.

sudo docker run -v ./license.txt:/root/kby-ai-idl/license.txt -p 7860:7860 -p 9005:9000 kbyai/idl:latest

SDK License

This Docker image utilizes KBY-AI's SDK, which requires a license for each machine or instance on which it runs.

• To get the machine code (HWID) for license, use the following syntax:

sudo docker run -e LICENSE="xxxxx" kbyai/idl:latest

• To request the license, please provide us with the machine code (HWID) obtained from your console.

Contact us:

🧙Email: contact@kby-ai.com
🧙Telegram: @kbyaisupport
🧙WhatsApp: +19092802609
🧙Discord: KBY-AI
🧙Teams: KBY-AI

Test API in KBY-AI Online Demo Page

You can test the SDK against static image here.

Test API with Gradio Demo

To test the Gradio demo from the Docker image, access it at http://127.0.0.1:7860 or http://localhost:7860.

API Instruction

API can be tested through Postman based on the following endpoint details.

1. Processing Image with File

• API endpoint: <base URL>/process_image
• Method: POST
• Content-type: multipart/form-data
• Request Parameters:

  Parameter	Type	Description
  image	File	ID document image file with *.jpg or *.png extension

• Request example
  

2. Processing with Base64-Encoded String

• API endpoint: <base URL>/process_image_base64
• Method: POST
• Content-type: application/json
• Request Parameters:
  

  Parameter	Type	Description
  base64	String	The base64-encoded string of the ID document image

• Request example
  

API Response Description in JSON

1. portraitReplace:

• Type: Float
• Description: A score indicating the likelihood that the portrait on the ID document has been tampered with or replaced. Higher values (closer to 1) suggest the portrait is likely original, while lower values may indicate potential tampering.
  

2. printedCopy:

• Type: Float
• Description: A score that assesses whether the document is a printed copy rather than an original document. Values close to zero suggest the document might be a printed copy, while higher values indicate originality.
  

3. screenReply:

• Type: Float
• Description: A score that determines if the document was displayed on a screen rather than being a physical ID. Lower values (closer to zero) imply a higher likelihood that the document is displayed on a screen.
  

4. Result status:

• Type: String
• Description: Indicates the status of the liveness check on the document. "Ok" typically means that the document passed initial verification. Other statuses could indicate different levels of authenticity or potential issues with the document.
  
• Status: Ok, Too close to camera!, Too close to border!, Document cropped!, Too small!, Multiple documents, Is Colorless!, Document not found!, Unknown
• Resonpse Example:

{
    "result": {
        "portraitReplace": 0.590566158294678,
        "printedCopy": 2.0435e-11,
        "screenReply": 0.002031173091382,
        "status": "Ok"
    },
    "resultCode": "Ok"
}