본문 바로가기 주메뉴 바로가기

(주)엠제이비전테크

Industrial Safety

Development background

An industrial safety vision system that applies deep learning video analytics to industrial machine vision, recognizing workers' bodies, PPE and positions in real time and stopping equipment at the moment of danger through direct PLC integration

Features

If CCTV monitoring watches 'the whole site', industrial safety machine vision attaches to each hazardous machine — presses, robots, conveyors — watching workers' hands and bodies millisecond by millisecond and linking directly to equipment control.

Machine-Mounted Safety Vision

  • CONVENTIONAL

    Sensor · Photoelectric Guards

    • Light curtains and area sensors are fixed in place, leaving blind spots at the sides and below
    • Protection is lost when bypassed for convenience
    • Cannot judge 'the person's condition', such as PPE use or working posture
    • No data for analyzing causes after an accident
  • MACHINE VISION

    Machine Vision Safety System

    • One camera monitors the front and sides of equipment in 3D, with freely configurable virtual safety zones
    • Video-based judgment cannot be bypassed, and any bypass attempt is itself logged as an event
    • Deep learning + rule-based judgment recognizes body parts, PPE, posture and movement at the same time
    • Video before and after hazard events is saved automatically → material for prevention training and reporting
  • 01

    Hazard Zone Intrusion Detection · Interlocked Stop

    Virtual Safety Zones are set around presses, robot cells and conveyor drives. When a worker's hand, arm or body enters a zone, a PLC digital output immediately sends a stop or slow-down signal and warning lights and buzzers alert the site. Multi-stage zones — warn → slow → stop — reduce unnecessary line stoppages.

    Hazard zone intrusion detection

  • 02

    Automatic PPE Compliance Check

    Automatically checks helmets, safety glasses, gloves, dust masks and harnesses when workers enter a work zone and during work. If PPE is missing, equipment start-up is blocked by interlock or managers are alerted, and compliance statistics are provided daily, weekly and monthly.

    Automatic PPE compliance check

  • 03

    Forklift · Heavy Load Collision Prevention

    Recognizes pedestrians in the path of forklifts, AGVs and cranes and warns both drivers and pedestrians by distance stage. Both vehicle-mounted (360° camera) and fixed aisle configurations are supported, applied first at collision hot spots such as intersections and doorways.

    Forklift · Heavy Load Collision Prevention

  • 04

    Abnormal Situation Detection (Falls · Fire · Leaks)

    Video detects falls and prolonged inactivity in lone-work areas, fire and smoke around equipment, and pipe leaks, and immediately notifies the control room and site supervisors. Adding thermal cameras also lets you monitor equipment overheating and abnormal electrical panel temperatures.

    Abnormal situation detection

Function

A 3-tier architecture of industrial cameras + edge AI controller + PLC integration. Connects with just 2 I/O points and no modification to existing equipment.

  • Industrial CameraGigE · 2~5MP · 60fps
  • Lighting/Thermal (Option)Low-light · backlight correction
  • Existing CCTV VideoRTSP input
Edge AI Controller On-board NPU, runs standalone on site Deep learning + rule-based judgment Event video storage
  • PLC / Safety RelayStop · slow · interlock
  • Warning Light / BuzzerInstant on-site alarm
  • Monitoring SW / MobileStats · reports · alerts
  • Abnormal Situation Detection (Falls · Fire · Leaks)

    New industrial cameras or existing CCTV via RTSP. Monitors abnormal temperatures when combined with thermal imaging.

  • Edge AI Judgment

    Analyzed directly on the site controller without a server. Safety functions keep working even if the network goes down.

  • Equipment Control Output

    Sends stop/slow/interlock signals via PLC digital I/O, Modbus TCP and safety relays.

  • Alarm · Logging

    Warning lights, buzzers and mobile push alerts fire together. Video before and after events and judgment logs are stored automatically.

  • Virtual Safety Zone Editor

    Draw warn/slow/stop zones as polygons on screen and save them per machine. Zones switch by work mode (run/setup/cleaning).

    OUT · DO1 WARN / DO2 STOP

  • PPE Compliance Judgment

    5 types: helmet, safety glasses, gloves, mask, harness. Retrained on color and shape to match the PPE issued at each site.

    OUT · INTERLOCK / REPORT

  • Moving Vehicle Proximity Alarm

    Calculates the distance between forklifts, AGVs or cranes and pedestrians, with 2-stage distance alarms and speed-limit signal output.

    OUT · SPEED LIMIT / HORN

  • Fall · Inactivity Detection

    Detects a fallen posture or no movement beyond a set time in lone-work areas and notifies immediately.

    OUT · ALERT / CALL

  • Fire · Smoke · Leak Detection

    Combines flame/smoke video detection with thermal temperature thresholds to minimize false alarms. Detects pipe and tank leaks on video.

    OUT · FIRE ALARM I/F

  • Safety Statistics · Reports

    Dashboards for hazard event counts, PPE compliance rates and stop history per machine. Safety & health management reports are generated automatically.

    OUT · PDF / EXCEL / API

Applications

Applications
Site Key Hazards Functions Applied Integrated Equipment
Press · Injection · Cutting Hands caught between dies, approach during slide descent Safety zonePPE Safety relay, two-hand control circuit
Robot automation cell Entry after opening the fence, cobot collisions Safety zoneInactivity Robot controller safety input
Conveyor · Distribution center Entanglement in drives, forklift-pedestrian collisions Proximity alarmSafety zone Conveyor inverter, forklift alarm
Chemical · Hazardous materials Leaks, fire, lone work in confined spaces Leak/FireCollapsePPE Fire alarm panel, PA system
Steel · Casting · High-heat equipment Approaching hot zones, passing under cranes ThermalProximity alarm Crane control panel, warning lights
상단으로