Notifications
Back

METSAKUUR Vision AI Technology

World-class Vision AI, validated through rigorous independent testing

Technology

Facial Recognition AI Technology

World-class Vision AI, validated through rigorous independent testing

Technology

· Vision AI Technology 

Technology — In-House Vision AI (Preview)
Real-time Face Recognition

From Engine to Middleware
100% in-house Vision AI
from Metsakuur.

We built every layer ourselves — the AI engine, the middleware, the encryption. Face recognition, anti-spoofing, behavior analysis and vital sign sensing run on one multimodal engine, with no third-party dependencies anywhere in the stack.

Accuracy 100%
Matching speed 0.06 sec
Users 10M+

· Facial Recognition Flow

How Face Recognition Works - From Capture

to Verification

Technology — Face Recognition Pipeline (Preview)
1
Capture

Captures live video from the camera and automatically corrects for lighting and angle.

2
Face Detection

Isolates the face region from the frame, handling multiple subjects at once.

3
Liveness Detection

Screens photo, video and mask attacks in real time to confirm a live subject.

4
Feature Extraction

Encodes distinctive facial features into a 512-dimensional vector for comparison.

5
Matching

Returns a 1:1 or 1:N result via cosine similarity in under 0.06 seconds.

· Key Features

Extracting and analyzing

facial landmarks

from video and images

Face Recognition

Analyze unique facial features
to identify users quickly and accurately.

The system extracts distinctive facial features and compares them with enrolled data to verify whether the person is a match. It can be applied to a wide range of services that require reliable user identification, including remote identity verification, access authentication, and membership recognition.

Face Detection Feature Encoding Identity Matching
Face recognition technology analyzing facial landmarks and matching results
Gender & Age Estimation

Analyze visual facial characteristics
to understand user demographics.

By analyzing visual patterns in facial images, the system estimates gender and age range. It provides demographic insights without directly identifying individuals, enabling data-driven applications such as visitor analytics, customer insights, and content optimization.

Gender Estimation Age Estimation Audience Analytics
Technology estimating gender and age characteristics from facial features
Behavior Recognition

Analyze human movement and flow
to recognize behaviors and situations.

The system analyzes posture, movement, and motion patterns across video sequences to recognize a variety of human behaviors. It supports intelligent video analytics for safer and more efficient spaces, including abnormal behavior detection, crowd analysis, and movement-flow monitoring.

Pose Estimation Action Recognition Flow Analysis
Behavior recognition technology analyzing human posture and movement patterns
Object Recognition

Detect and recognize a wide range of objects
and classify them accurately.

The system detects and classifies multiple objects in images and video. By analyzing the location and movement of people, vehicles, products, facilities, and other objects, it can be applied across smart cities, retail, security, industrial environments, and many other fields.

Object Detection Classification Tracking
Object recognition technology detecting and classifying people, vehicles, and other objects
Liveness Detection & Anti-Spoofing

Verify that the user is physically present
and block fraudulent authentication attempts.

The system determines whether a face captured by a live camera belongs to a real, physically present user and detects spoofing attempts using photos, videos, or screen replays. By adding liveness verification to face recognition, it strengthens the reliability of remote identity verification and face-based authentication services.

Liveness Detection Anti-Spoofing Presentation Attack Detection
Liveness detection technology comparing a genuine user with a photo-based spoofing attempt

· METSAKUUR Vision AI Technology 

Technology — Vision AI Built In-House

From Engine to Middleware
100% in-house Vision AI
from Metsakuur.

We built every layer ourselves — the AI engine, the middleware, the encryption. Face recognition, anti-spoofing, behavior analysis, and vital sign sensing run on one multimodal engine, with no third-party dependencies anywhere in the stack.

Accuracy 100%
Matching speed 0.06 sec
Users 10M+
Real-time Face Recognition

· Facial Recognition Flow 

How Face Recognition Works - From Capture to Verification

Technology — Face Recognition Pipeline
Capture

Captures live video from the camera and automatically corrects for lighting and angle.

Face Detection

Isolates the face region from the frame, handling multiple subjects at once.

Liveness Detection

Screens photo, video, and mask attacks in real time to confirm a live subject.

Feature Extraction

Encodes distinctive facial features into a 512-dimensional vector for comparison.

Matching

Returns a 1:1 or 1:N result via cosine similarity in under 0.06 seconds.

· Vision AI Technology 

Extracting and analyzing facial landmarks from video and images

Face Recognition

Analyze unique facial features
to identify users quickly and accurately.

The system extracts distinctive facial features and compares them with enrolled data to verify whether the person is a match. It can be applied to a wide range of services that require reliable user identification, including remote identity verification, access authentication, and membership recognition.

Face Detection Feature Encoding Identity Matching
Face recognition technology analyzing facial landmarks and matching results
Gender & Age Estimation

Analyze visual facial characteristics
to understand user demographics.

By analyzing visual patterns in facial images, the system estimates gender and age range. It provides demographic insights without directly identifying individuals, enabling data-driven applications such as visitor analytics, customer insights, and content optimization.

Gender Estimation Age Estimation Audience Analytics
Technology estimating gender and age characteristics from facial features
Behavior Recognition

Analyze human movement and flow
to recognize behaviors and situations.

The system analyzes posture, movement, and motion patterns across video sequences to recognize a variety of human behaviors. It supports intelligent video analytics for safer and more efficient spaces, including abnormal behavior detection, crowd analysis, and movement-flow monitoring.

Pose Estimation Action Recognition Flow Analysis
Behavior recognition technology analyzing human posture and movement patterns
Object Recognition

Detect and recognize a wide range of objects
and classify them accurately.

The system detects and classifies multiple objects in images and video. By analyzing the location and movement of people, vehicles, products, facilities, and other objects, it can be applied across smart cities, retail, security, industrial environments, and many other fields.

Object Detection Classification Tracking
Object recognition technology detecting and classifying people, vehicles, and other objects
Liveness Detection & Anti-Spoofing

Verify that the user is physically present
and block fraudulent authentication attempts.

The system determines whether a face captured by a live camera belongs to a real, physically present user and detects spoofing attempts using photos, videos, or screen replays. By adding liveness verification to face recognition, it strengthens the reliability of remote identity verification and face-based authentication services.

Liveness Detection Anti-Spoofing Presentation Attack Detection
Liveness detection technology comparing a genuine user with a photo-based spoofing attempt
mobile background

· How Safe Facial Recognition Technology

How Secure Is Facial Recognition Technology?

Anti-Spoofing
Distributed Biometric Data Management
E2E Encryption Technology
mobile background
Face Detection
Fake Check 1
Face Area Storage
＋
Preprocessing & Resizing
Fake Check 2
Feature Extraction on Server
Comparison with Stored Data

  1  

AI-Based Anti-Spoofing Technology

AI trained with over 20,000 Fake–Real datasets Development of 5-layer Anti-Spoofing defense against fraudulent authentication attempts

  2  

RGB-Based Fake Detection

Distinguishes photos, videos, and spoofed biometrics from real faces Provides strong fake detection on devices equipped with standard RGB cameras
mobile background
Feature Extraction
Feature Splitting
METSA FR
METSA FR
KFTC Server
Split Feature Storage
Feature Recombination
Score
0.99
Face Authentication

 1 

Distributed Security for Financial Authentication

First in Korea to pass the KFTC Distributed Biometric Information Management Performance Evaluation, ensuring secure protection and management of biometric data.

 2 

Standards-Based Distributed Authentication

Data splitting and merging based on distributed management standards, with a system architecture designed to support distributed authentication.
mobile background
Encryption / Decryption
Client Module
WAS Server
Relay Module
Face Authentication Server
Encryption / Decryption
Server Module

 1 

End-to-End Encrypted Communication

Communication between the data sender and receiver is encrypted, preventing access from intermediate servers or external parties.

 2 

TOTP-based Dynamic Encryption

A module applying a TOTP-based algorithm changes the encryption key at regular intervals, making the encryption extremely difficult to predict or compromise.

· Cross-Platform Support

Deployment Options for Every Environment

Mobile & PC Support

AI Vision capabilities can be deployed across the device environments customers actually use, from smartphones to PCs.

✓Supports Android- and iOS-based mobile services
✓Integrates with Windows- and Linux-based PCs and systems

IoT & Edge Device Support

From compact on-site devices to high-performance computing systems, deployment can be scaled to suit service environments and processing requirements.

✓Integrates with access terminals, kiosks, and camera-based devices
✓Supports both on-device and server-side processing architectures

On-Premise & Cloud Integration

Flexible deployment options enable integration with internal systems or cloud environments based on each organization’s security policies and infrastructure.

✓Supports deployments centered on internal enterprise servers
✓Integrates with cloud-based service architectures
Supported Environments
Android iOS Windows Linux Mobile PC IoT Device Edge Device On-Premise Cloud
FLEXIBLE INTEGRATION Your environment does not adapt to our technology.
Our technology adapts to your environment.

We design flexible integration options around your existing service architecture and operating environment.

The Value of Cross-Platform Support

By avoiding dependency on a single device or operating system, cross-platform support reduces the need to rebuild an entire system as services scale. The same AI Vision technology can be deployed consistently across mobile devices, on-site equipment, servers, and other environments.

What industry applied to

Where Vision AI

Is Already at Work

Financial Industry

Identity verification for financial services
and enhanced security

Public Institutions

Personal identity verification
and safety & welfare

Construction Sites

Workforce management for construction sites
and enhanced safety & security

Education Services

Remote online exam
monitoring service

Access & Attendance Management

An efficient attendance management system
for large-scale personnel

Smart Retail

Customer marketing
and product recommendations

mobile background

· Cross-Platform Support

Deployment Options for

Every Environment

Mobile & PC Support

AI Vision capabilities can be deployed across the device environments customers actually use, from smartphones to PCs.

✓Supports Android- and iOS-based mobile services
✓Integrates with Windows- and Linux-based PCs and systems

IoT & Edge Device Support

From compact on-site devices to high-performance computing systems, deployment can be scaled to suit service environments and processing requirements.

✓Integrates with access terminals, kiosks, and camera-based devices
✓Supports both on-device and server-side processing architectures

On-Premise & Cloud Integration

Flexible deployment options enable integration with internal systems or cloud environments based on each organization’s security policies and infrastructure.

✓Supports deployments centered on internal enterprise servers
✓Integrates with cloud-based service architectures
Supported Environments
Android iOS Windows Linux Mobile PC IoT Device Edge Device On-Premise Cloud
FLEXIBLE INTEGRATION Your environment does not adapt to our technology.
Our technology adapts to your environment.

We design flexible integration options around your existing service architecture and operating environment.

The Value of Cross-Platform Support

By avoiding dependency on a single device or operating system, cross-platform support reduces the need to rebuild an entire system as services scale. The same AI Vision technology can be deployed consistently across mobile devices, on-site equipment, servers, and other environments.

What industry applied to

Where Vision AI

Is Already at Work

Financial Industry 

Identity verification for 
financial services and enhanced security

Education Services 

Remote online exam
monitoring service 

Public Institutions 

Personal identity verification
and safety & welfare 

Access & Attendance Management 

An efficient attendance
management system
for large-scale personnel 

Construction Sites 

Workforce management for
construction sites and enhanced safety & security

Smart Retail 

Customer marketing 
and product recommendations 

Start Your Business Transformation with Vision AI

From finance and government to healthcare, education, and industrial sites, Metsakuur delivers secure, frictionless solutions for your business.

Start Your Business Transformation

with Vision AI

From finance and government to healthcare, education, and industrial sites,

Metsakuur delivers secure, frictionless solutions for your business.

Company
Metsakuur Company
CEO
Jihoon Lee
Address
15-3, Eonju-ro 136-gil, Gangnam-gu, Seoul, Republic of Korea
Tel
+82-2-6743-0011
Email
master@metsakuur.com
Business Registration NO.

179-88-01680

Family Sites
+
iJANUS OTM
TOP

Copyright© METSAKUUR All Rights Reserved.

Company
Metsakuur Company
CEO
Jihoon Lee
Address
15-3, Eonju-ro 136-gil, Gangnam-gu, Seoul, Republic of Korea
Tel
+82-2-6743-0011
Email
master@metsakuur.com
Business 
Registration NO.

179-88-01680



Family Sites
+
iJANUS OTM

Copyright© METSAKUUR All Rights Reserved.