하이트론씨스템즈

SOLUTIONS Edge-based AI engine with real-time object
classification features for intelligent video solutions

AI Industrial Safety

Overview

By utilizing various input devices (CCTV, drones, body cameras, etc.) at industrial sites,

Video information is input and analyzed based on AI machine learning/deep learning.
This allows the system to automatically identify not only objects at the site but
also their dangerous situations or conditions in real time.

It then provides managers with alerts on hazardous situations and offers workers corresponding actions to take.
This is a disaster safety management system designed specifically for industrial sites.

Equipped with an AI engine that recognizes hazardous situations and
conditions through optimized labeling data, the solution enables real-time management of
both structured and unstructured objects (e.g., tools, and work environments) at construction sites.

Categories Requirements Implementation AI-related Technologies Remarks
Worker Protection Detection of insulated protective gear and hardhat usage ● Object condition recognition Logistics Center safety management
Detection of chemical protective gear usage ● Object condition recognition
Hardhat detection for robotic facilities ● Object condition recognition + Object situation recognition
Worker detection
(Fall detection, no movement for a certain time)		 ◐
(In progress)		 Object situation recognition, hazardous pattern analysis
Access Control Detection when entering from an unauthorized route ● Object recognition + Hazard zone setting
Entry and access alerts during equipment operation ● Object recognition + Hazard zone setting Equipment operation status assessment
1) Data learning required
2) Sensor integration
Alert when entering without unlocking interlocks in robotic facilities ● Object recognition + Hazard zone setting
Detection of entry into restricted areas
(1st and 2nd warning when within 90cm/50cm of the charging area)		 ● Object recognition + Hazard zone setting
Facility Management Facility detection : Deformation, collapse, and tipping detection ○
(In learning)		 Object condition recognition
Earthquake detection: Vibration, collapse detection ○
(In learning)		 Object recognition (collapse)

AI (Artificial Intelligence) Video Analysis
Object Condition Identification Technology

  • Object Recognition Technology
    Structured/Unstructured object recognition
    Unstructured object recognition using polygon labeling techniques
    (*Structured objects have fixed shapes, while unstructured objects have defined characteristics but can vary in form.)

    Development of AI Engine Through
    Innovative Data Labeling Techniques

  • Situation/Condition Identification Technology
    Analysis of object situations/conditions (hazardous states)
    Using AI video integration technology, objects are
    initially recognized, and in the second step,
    AI object classification techniques are applied to
    identify the characteristics of the recognized objects

    (Example) In industrial settings,
    the characteristics of objects used in the
    work environment are governed by regulations
    under the Occupational Safety and Health Act

    Safety & Hazard

go top