Human Detection AI System

Human Detection System AI is a professional industrial safety solution designed for continuous monitoring of restricted, robotic and machine-operating areas.

The system uses AI-based human detection together with configurable Main ROI, AI Ignore Zones and Dynamic ROI zones to determine whether a person is present inside a safety-controlled area. It is designed for local industrial deployment and can integrate with a client company PLC, machine control system or robot control architecture for automated alarm and safety-response workflows.

The platform provides a centralized operator interface for live AI monitoring, clean CCTV viewing, ROI configuration, event logs, storage management, system health, user management and company branding.

The solution is suitable for factories and production environments where conventional fixed safety zones are not sufficient and the monitored area may need to change according to machine status, PLC inputs or production conditions.

Key Features of the Human Detection AI System

Real-Time AI Human Detection

The system continuously processes camera streams and identifies valid human presence inside configured safety areas.

Detection sensitivity, AI device selection, inference rate, confidence, confirmation time and image size can be configured according to the deployment requirements. The supplied configuration supports GPU processing with CPU fallback.

Multi-Camera Industrial Monitoring

The supplied system configuration supports up to five camera channels, with individual camera settings, RTSP/USB source configuration, FPS settings and optional lower-resolution AI sub-streams.

Main ROI Safety Zone

A permanent Main ROI can be defined for every camera to ensure AI decisions are limited to the actual operational or restricted area.

The system evaluates the detected person's bottom-center / feet position, helping the ROI correspond more accurately to the physical floor area.

AI Ignore Zones

Selected areas inside the Main ROI can be excluded from detection logic using AI Ignore polygons.

The system supports up to 5 AI Ignore polygons per camera in the supplied configuration.

Dynamic ROI Control

Dynamic ROI allows safety zones to become active or inactive according to PLC signals or commissioning simulation inputs.

The system supports up to 10 globally numbered Dynamic ROI polygons across the five-camera system.

When a Dynamic ROI is OFF, that area remains reserved/inactive. When the corresponding input turns ON, the same polygon becomes an active human-detection area.

Siemens PLC Integration

The application can communicate with a Siemens S7-1200 PLC using Snap7-based communication.

The supplied configuration provides separate signals for:

• Human Detected
• System Fault
• System OK
• Python Heartbeat

PLC output and Dynamic ROI input mapping are independently configurable according to the commissioned client control architecture.

PLC-Controlled Dynamic Safety Zones

Dynamic ROI states can be received from PLC data-block bits.

In production mode, ROI input status is read from the PLC and shown as read-only status indicators in the application.

Continuous Detection Confirmation

To reduce brief or false triggers, the system can require a person to remain continuously detected for a configurable period before generating a confirmed event.

For example, with a 3-second confirmation time, short detections do not trigger the output; a continuously valid detection reaching the configured time becomes a confirmed human event.

Automatic Alarm Logic

In normal production operation, AUTO MODE allows confirmed human detections and configured safety logic to determine the alarm/output condition.

Manual Signal ON and password-protected Bypass modes are also available for commissioning and authorized maintenance activities.

Event Snapshot & Video Recording

Confirmed detections can automatically create:

• Detection snapshots
• Event video clips
• Timestamped event records
• Camera identification

Operators can review historical events from the Event Logs page.

Intelligent Storage Management

The system provides configurable storage overwrite management.

When disk usage reaches the selected threshold, the storage manager can use another configured storage destination or remove the oldest HESHAM AI event files first to free space.

System Health Monitoring

The health dashboard provides visibility into:

• CPU usage
• RAM usage
• GPU status
• Disk condition
• Temperature
• Individual camera status

This helps maintenance teams identify degradation before it becomes a production interruption.

Camera Health & Automatic Recovery

Each camera can display operational states such as:

• ONLINE
• CONNECTING
• DOWN / FAULT
• Live FPS
• AI Ready
• Human Detected

This provides quick fault identification for camera, network or AI-processing issues.

Clean CCTV Live View

A dedicated CCTV page provides camera streams without AI overlays, allowing engineers to distinguish between a camera/network issue and an AI/ROI configuration issue.

Role-Based User Security

The system supports individual users and administrator functions including:

• User creation
• Operator/Admin roles
• Password changes
• Administrator password reset
• Account status review
• Last-login information
• Removal of obsolete accounts

Custom Client Branding

Administrators can configure the system with the client company's name and logo while maintaining HESHAM system identity.

Branding does not affect AI, PLC, ROI or storage operation.

Automatic Windows Service Operation

The monitoring backend is designed to run as a Windows service, allowing the application to start automatically when the industrial PC powers on.

This reduces operator dependency and is suitable for continuous production use.

Typical System Architecture

Industrial Cameras → AI Processing Computer → Detection & Decision Logic → Siemens PLC → Client PLC / Robot Control System

The AI software receives camera streams, processes human detection and ROI conditions, then communicates required status or alarm outputs through the commissioned PLC architecture.

This allows the solution to become part of a broader industrial machine-safety and automation system. The manual also makes clear that certified emergency-stop, guarding and fail-safe circuits must remain within the approved plant safety architecture.

Main Software Modules

AI Monitor

Live AI streams, detection status, FPS and alarm controls.

CCTV Live View

Clean video streams without AI overlays.

Main ROI Setup

Permanent safety detection area and AI Ignore zones.

Dynamic ROI Setup

PLC-controlled dynamic safety polygons.

Dynamic ROI Control

Simulation and Siemens PLC input management.

Event Logs

Detection history, snapshots and event recordings.

Storage Manager

Automatic overwrite threshold and storage-drive management.

Settings & Health

Camera configuration, AI configuration, PLC output settings and system health monitoring.

Company Branding

Client identity and logo configuration.

User Security

User accounts and password management.

Technical Highlights

The exact production parameters are commissioned per site; the manual identifies values such as 0.55 confidence, 3-second confirmation, 640 AI image size, 12 FPS GPU target and 100 ms Dynamic ROI polling as supplied baseline guidance rather than universal fixed requirements.

Industrial Applications

Suitable for:

• Robotic Welding Areas
• Automotive Manufacturing
• Assembly Lines
• Machine Safety Zones
• Restricted Production Areas
• Automated Manufacturing Cells
• Industrial Robot Work Cells
• Material Handling Areas
• Smart Factory Safety Monitoring
• Human–Machine Interaction Zones
• Industrial AI Surveillance
• Dynamic Machine Safety Applications

Benefits for Industrial Safety

Improved Personnel Safety

Provides continuous AI-based monitoring of areas where human entry must be controlled.

Reduced False Triggers

ROI filtering, ignore polygons and configurable confirmation timing help prevent short or irrelevant detections from becoming confirmed events.

Flexible Safety Zone Control

Dynamic ROI makes it possible for detection areas to change according to PLC or machine status.

Centralized Monitoring

Operators can monitor AI, cameras, alarms, system status, events and storage from a single application.

Industrial Integration

Designed to connect AI detection with PLC and automation systems.

Local Data Processing

Configuration, event database, snapshots, event clips and service logs are stored locally in the supplied architecture.

Production-Oriented Reliability

The system includes camera health, storage management, automatic service startup and system-health monitoring.

System Highlights

AI Human Detection

Real-time person detection for industrial safety zones.

Dynamic ROI

PLC-controlled active detection zones.

Multi-Camera Monitoring

Centralized AI and CCTV camera supervision.

PLC Integration

Communication with industrial Siemens PLC architecture.

Event Evidence

Automatic snapshots and event video recording.

System Health

Continuous monitoring of hardware and camera health.

Automatic Storage

Intelligent local storage and overwrite management.

Secure Access

Role-based user and administrator management.

Frequently Asked Questions

How does the Human Detection AI System work?

Industrial cameras feed AI person detection. Main ROI and dynamic ROI zones determine relevant detections; confirmed events can generate site-configured PLC signals and event records.

Can the human detection system integrate with a Siemens PLC?

The system supports configurable Siemens S7-1200 communication. Final I/O mapping and machine-control responses are verified during commissioning.

Does AI human detection replace certified machine guarding?

No. AI monitoring is an additional safeguard. Required emergency stops, interlocks and certified guarding remain part of the approved safety system.