AI Vision

The cameras you already have
become industrial AI vision, as is

With no dedicated AI camera or costly edge box, connect your existing standard IP cameras directly to detect people, vehicles, protective-equipment use, and fire in real time. Implement industrial AI vision at a reasonable cost by combining a dedicated PC program with EGNOX Cloud.

  • Existing IP cameras (RTSP)
  • PC + cloud
  • PPE & fire detection
  • YOLO object detection
  • Up to 8 channels
  • Reasonable cost
// LIVE
In action

View up to 8 sites on a single screen,
in real time

Display feeds from multiple cameras at once on the PC screen, and review the AI analysis results together on the same screen.

EGNOX Vision Agent — PC program screen detecting people, helmets, and safety vests in real time from construction-site footage

EGNOX Vision Agent — detecting person, helmet, and vest in real time from actual construction-site footage, connected to the cloud via MQTT. This is a live screen in operation.

Up to 8 channelsSimultaneous monitoring
on a single PC
RTSPConnect your existing
IP cameras as is
PC + cloudImplemented at reasonable cost
with no dedicated AI hardware
Model swapExpand detection items
(PPE, fire, quality)
// 01 — PAIN POINTS
Pain points

These situations prompt
teams to consider adoption

You have CCTV, but it only records

The cameras are installed, but they are only used to look back at footage when something goes wrong. It is a setup where you only check after an incident occurs.

People manually check protective-equipment use

People patrol or watch monitors to visually verify whether helmets and safety vests are worn. The larger the site, the more areas get missed.

You learn about fires and incidents too late

Even when a fire or hazardous situation occurs, you only find out through the reporting chain. Early-response time grows accordingly.

Dedicated hardware costs are a burden

Adopting AI vision means paying for dedicated AI cameras or costly edge boxes. Getting started itself is daunting.

// 02 — OVERVIEW
Product overview

Use the cameras you already have,
and keep costs down

EGNOX Vision is an industrial AI vision solution that links a dedicated PC desktop program with EGNOX Cloud (EGNOX Platform).

🎥
Existing IP cameras, as isConnect standard, low-cost IP cameras (RTSP) directly. No dedicated AI camera or costly edge box is needed.
♻️
Reuse installed CCTVYou can reuse CCTV that is already installed, keeping the adoption burden low.
🖥️
Processed directly on the on-site PCVideo analysis runs on the on-site PC, and only the necessary data is sent to the cloud.
💰
PC + cloud combinationImplement and extend AI capabilities efficiently with a PC program and cloud combination, without costly dedicated AI hardware.
There are two key points. One, perform a range of AI detection with the IP cameras you already have. Two, implement AI capabilities at a reasonable cost with a PC program and the cloud.
EGNOX Platform cloud dashboard — screen for camera monitoring and detection-data lookup

EGNOX Platform — review and monitor detection results, snapshots, and data sent from the PC on the cloud dashboard. This is a real screen.

// 03 — HOW IT WORKS
Camera to cloud

From camera to cloud,
it works in four steps

1

Connect the camera

Connect your existing IP cameras to the PC program (Qt) via RTSP. Monitor up to 8 cameras at the same time.

2

Real-time AI analysis on the PC

The PC displays live video while YOLO-based AI detects objects. Recording, digital zoom, and image processing are also supported.

3

Cloud integration

Send detection metric data, snapshot images, alarm events (fire, PPE violations), and agent status to EGNOX Platform via MQTT.

4

Dashboard, alarms, analysis

From the cloud, it carries through to dashboard review, alarm reception, data analysis, and video/event archiving.

// 04 — FEATURES
Detection

What it detects

Automatically detect protective equipment, fire, and people, and expand detection items by swapping the AI model.

People & vehicle detection

Used for access and headcount monitoring and for verifying entry into hazardous zones.

Automatic check of protective-equipment (PPE) use

Alerts immediately when helmets, safety vests, safety shoes, and the like are not worn.

Early fire & smoke detection

Quickly recognizes fire and smoke to rapidly grasp conditions before and after an incident.

Quality & custom detection

Simply swapping the AI model extends to new detection items such as quality inspection.

Detection items are not fixed. Just by swapping the AI model, the system extends to new detection items.
Proof of concept complete — we confirmed real-time detection of people, helmets, and safety vests from actual construction and industrial site footage, with snapshots and data sent to EGNOX Platform.
// 05 — WHY IT MATTERS
Cost & adoption

Start with AI vision
without buying new hardware

🎥
Reuse existing IP camerasImplement industrial AI vision at a reasonable cost, without purchasing dedicated AI cameras or edge boxes.
💰
PC + cloud combinationImplement and extend AI capabilities with just a PC program and the cloud, without costly dedicated AI hardware.
🔒
Local-processing architectureVideo analysis runs on the on-site PC, and only the necessary metrics, snapshots, and events are sent to the cloud.
📈
Scale as neededStart with one camera and scale up to 8, and broaden detection items by swapping the model.
// 06 — USE CASES
Use cases

Where and how
it is used on the floor

These are example use scenarios you can configure at deployment. The actual configuration may vary depending on site conditions.

🦺 Industrial safety

Real-time detection and alerts for missing protective equipment

Detects workers without helmets or safety vests in real time and immediately alerts the responsible staff.

Instant alert on non-complianceReduced reliance on patrols
🔥 Early fire detection

Rapid fire and smoke recognition for early response

Quickly recognizes fire and smoke to support early response.

Early recognitionAutomatic event logging
🚶 Access & headcount monitoring

Check access status through people and vehicle detection

Detects people and vehicles to check access and headcount status.

Headcount statusHazardous-zone entry
🔍 Quality monitoring

Detect quality anomalies and defects via model swap

Extend to quality-anomaly and defect detection by swapping the AI model.

Model-swap expansionDefect detection
// 07 — SPEC
Tech specs

Tech specs

CameraCompatible with standard, low-cost IP cameras (RTSP); no dedicated AI camera required; up to 8 cameras monitored simultaneously
PC programQt-based desktop program — live video display, recording, digital zoom, and image processing
AI detectionYOLO-based object detection; expand detection items by swapping the model
Cloud integrationSent to EGNOX Platform via MQTT (detection metric data, snapshots, alarm events, agent status)
Cloud featuresDashboard, alarms, data analysis, archiving
Detailed policyRetention periods, data-transmission intervals, and similar terms are agreed based on plan and requirements
// 08 — EDGE COMPUTE
PC or edge

On a PC, or on edge computing

Start floor safety monitoring lightly with PC + cloud, and when on-device real-time processing is needed for cases like line integration or precision inspection, run it on NVIDIA Jetson edge computing.

NVIDIA Jetson Orin Nano development board
Entry

Jetson Orin Nano

Real-time object detection on a compact, low-power edge device. Suited to a single line or a small-scale site.

NVIDIA Jetson AGX Orin module
Standard

Jetson AGX Orin

A standard edge platform that runs real-time on-device inference on multi-camera, high-resolution video.

NVIDIA Jetson AGX Thor development kit
High-performance

Jetson AGX Thor

A high-performance platform that handles even large, demanding vision models at the edge.

* The edge-computing model is selected to fit site requirements such as camera count, resolution, and detection items.

Development complete

EGNOX Vision PCB
precision quality-inspection solution

Combines an industrial global shutter machine-vision camera with vision AI to automatically detect PCB defects in real time, without stopping the line.

SMT Edge AI Inspector — screen inspecting PCB components and placement status in real time on NVIDIA Jetson edge

SMT Edge AI Inspector — inspects PCB components and placement status in real time on NVIDIA Jetson edge computing. This is a real screen.

Why global shutter

PCB defects in real time,
without stopping the line

When a standard camera (rolling shutter) captures a fast-moving subject, the frame is recorded line by line with a time offset, so the image warps or smears. A global shutter records the entire frame at once, so you get a sharp, motion-blur-free image even while the line is moving.

    📸
    Image quality secured from the camera stageA sharp image is essential to accurately distinguish fine defects.
    📐
    Judgments paired with evidence imagesStandardize inspection criteria by judging from evidence images rather than human intuition.
EGNOX Vision PCB quality-analysis dashboard — screen showing first pass yield (FPY), DPPM, OK/NG trends, defect-type Pareto, per-component NG frequency, and LOT progress

PCB quality-analysis dashboard — aggregates first pass yield (FPY), DPPM, defect-type Pareto, per-component (RefDes) NG frequency, and LOT progress on one screen. This is a real screen (displayed figures are example data).

// PCB — FEATURES
Key features

Key features

Precision capture with a global shutter

Captures precisely based on an industrial global shutter machine-vision camera.

Motion-blur-free images

Secures sharp, motion-blur-free images even in high-speed-motion, precision-inspection environments.

Automatic AI defect detection and judgment

Vision AI analyzes the image to automatically detect and judge whether the PCB is defective.

Judgment-evidence image logging

Standardizes inspection criteria by retaining the evidence image for each judgment.

Expandable detection items

Address new defect types and product families by swapping the AI model.

EGNOX Platform integration

Centrally manage judgment data, snapshots, and events in the cloud.

// PCB — USE CASES
Use cases

On which floors

🏭
PCB manufacturing and placement linesAutomatically judges good vs. defective in real time on a moving line.
🔬
Precision component inspectionDistinguishes fine defects that the human eye easily misses, using consistent criteria.
📊
Quality-data accumulationCollect judgment results and evidence images in EGNOX Platform to analyze defect trends.
// PCB — SPEC
Tech specs

Tech specs

CameraIndustrial global shutter machine-vision camera
Detection methodAI-vision-based automatic defect detection and judgment
Development statusDevelopment complete
IntegrationEGNOX Platform (cloud) — transmits judgment data, snapshots, and events
Detection items & performanceDetectable defect types, precision, and processing speed (tact-time) vary by inspection target and line conditions, so they are agreed based on requirements
// COMPARE
Two solutions

How the two solutions differ

Same vision AI engine, different cameras. Both solutions share EGNOX Platform integration but use the camera suited to the purpose.

EGNOX VisionEGNOX Vision PCB
CameraReuse your existing standard IP cameras (RTSP)Industrial global shutter machine-vision camera
PurposeReal-time monitoring of floor safety, headcount, and firePCB precision quality inspection and defect detection
StrengthsCost-efficient adoption by reusing existing hardwareDetects fine defects with sharp, motion-blur-free images
In commonVision AI analysis + EGNOX Platform (dashboard, alarms, data) integration

Use EGNOX Vision for floor safety and headcount monitoring, and EGNOX Vision PCB for precision quality inspection that picks out even fine defects. Start with one product and expand as far as you need.

Next step

Get started right now,
with the cameras you have

You can start industrial AI vision at a reasonable cost with your existing IP cameras and a single PC. For a PCB inspection line, we review a detection setup matched to your line conditions together.

Contact sales