Directory

Discover detailed profiles of PoC companies.

SIJUNG Co., Ltd.

Hyper AI Vision Computer System for Serverless Real-Time Event Detection

Specialization

Edge AI Vision Computer
On-Device AI Surveillance
Real-Time Event Detection

Keywords

AI Vision Station
Serverless AI
Edge AI Camera

SIJUNG develops A.I. Vision Station, a serverless on-device AI vision computer for real-time event detection.
The system performs AI judgment directly at the camera edge, reducing dependence on centralized analysis servers.
It detects events such as fire, intrusion, accidents, crowd risks, weather visibility issues, and abnormal situations in real time.
The solution is designed for public safety, ports, airports, industrial sites, disaster prevention, and smart city infrastructure.

(1) Owned Technology

Serverless edge AI architecture for real-time video analytics.
On-device multi-task inference using a single backbone model.
NPU optimization technology using INT8 quantization and early-exit logic.
Real-time anomaly detection pipeline for public infrastructure.
Dynamic privacy masking for face and license plate de-identification before transmission.
Proprietary AI training dataset with more than 1 million field data samples from harsh environments.

(2) Applied Products

A.I. Vision Station
- On-device AI vision computer for real-time event detection.
- Operates without a high-cost centralized analysis server.
- Supports public safety, industrial monitoring, and disaster response.
Vision OS Framework
- Lightweight operating system optimized for edge inference.
- Supports memory allocation, heat control, and stable AI operation.
AI Model Update & Maintenance Service
- Recurring software, AI model update, and maintenance model for installed devices.

(1) Problem Definition / Development Background

CCTV installations are rapidly increasing, but human monitoring personnel cannot scale at the same speed.
The surveillance bottleneck is no longer camera quantity, but the lack of real-time judgment.
Conventional CCTV systems depend on centralized servers, causing high infrastructure cost, latency, bandwidth burden, and privacy risks.
Public safety markets are shifting from post-event recording to real-time detection and judgment.

(2) Value Proposition / Solution

SIJUNG distributes AI judgment to each camera, enabling real-time on-site event detection.
The solution removes the need for expensive centralized AI servers.
It reduces bandwidth use, power consumption, and deployment cost while improving scalability.
On-site de-identification before transmission improves privacy protection for public and industrial environments.

Serverless Architecture: Performs AI inference directly on the device without central server dependency.
Cost Advantage: Based on 16 units, the system can save approximately 70% of deployment budget compared with competitor systems.
Low Power Consumption: Reduces total power usage by around 60% compared with centralized server-type systems.
Privacy Protection: Enables hardware-level face and license plate masking before data transmission.
Fast Inference: Provides low-latency AI judgment under test conditions.
Proven Public References: More than 17 public-sector references across ports, airports, roads, forests, military, and local governments.

Secured Public Procurement Service Innovative Product designation.
Built public procurement fast-track channels through government innovation certification.
Delivered solutions to major public and infrastructure institutions.
Key references include Busan Port VTS, Uzbekistan Meteorological Service, Korea Hydrographic and Oceanographic Agency, East Sea Coast Guard, Incheon International Airport, KALIS, Korea Expressway Corporation, Korea Gas Corporation, Republic of Korea Navy, Daejeon Tourism Organization, and Korea Forest Service.
Initial addressable market includes more than 4,600 units across construction, forest service, maritime, meteorological, local government, and tourism sectors.

(1) PoC Progress

SIJUNG has delivered to the Uzbekistan Meteorological Service.
The company plans to convert PoC projects into full-scale commercial deployments through local partnerships.
Target global use cases include airport visibility monitoring, maritime safety, port security, wildfire detection, and toxic smoke detection.

(2) Global Expansion

SIJUNG aims to expand from domestic public procurement into global AI infrastructure markets.
Target verticals include airport and aviation, maritime and port, industrial safety, disaster prevention, and smart city monitoring.
The company plans to combine proprietary hardware, software licenses, and AI modules into long-term bundled service models.