SCENARIO Run all detection agents simultaneously
LIVE
SIMULATION
IDLE FPS: -- Stream: Idle

Live Video Feed

Camera 1

Click "Start Stream" to begin monitoring

Detection Status

Operator Status

--
0.0000
0
0.0 deg

Health Emergency

Normal
No
0%
0

Oil Leak Detection

Not Detected
0%
0

Accident Risk

Normal
0%
0.00
0.00
0

Active Alerts

0

No active alerts

Event History

Time Type Severity Message
No events recorded

Training Data Upload

Reference Image Analysis (AI Vision)

Upload a reference image and describe what you expect the AI to monitor. The system will analyze the scene and provide structured understanding.

💧

Oil Leakage Detection

Automated visual identification of unintended oil or hydraulic fluid spills, drips, or pooling on equipment surfaces, floors, or operational areas using computer vision.

What It Detects

  • Dark fluid patches forming on floor surfaces or equipment
  • Dripping or streaming fluid from machinery, pipes, or joints
  • Gradual spreading of wet regions over time
  • Pooling of fluids in low-lying areas near equipment

How It Works

  • HSV color space analysis for dark/brownish fluid regions
  • Contour detection for irregular fluid boundaries
  • Area thresholding to filter noise
  • OpenAI Vision API for contextual confirmation

Severity Levels

LowSmall drip, area < 500px MediumVisible leak, area 500-2000px HighSignificant spread, area 2000-5000px CriticalMajor spill, area > 5000px
🧑

Operator Status Monitoring

Real-time tracking and classification of a human operator's posture, position, and movement using AI pose estimation. Tracks nose, shoulders, and hips to determine stability.

Classifications

Stable

Controlled movement, body landmarks maintain expected positions with minimal displacement between frames.

Unstable

Erratic or abnormal movement patterns. May indicate stumbling, falling, dizziness, fatigue, or medical emergency.

Key Metrics

  • Movement Score: Average displacement of body landmarks (0.0 = still)
  • Instability Count: Consecutive frames exceeding threshold
  • Pose Keypoints: 3D coordinates for nose, shoulders, hips
⚠️

Deviation Detection

Identification of any condition, event, or behavior that departs from the established normal baseline. The umbrella concept encompassing operator instability and equipment anomalies.

Types of Deviation

Movement Deviation

Operator movement exceeds normal baseline: sudden jerking, falling, swaying, running in no-run zone.

Visual Deviation

Unexpected visual elements: oil puddles, smoke, fire, debris, equipment state changes.

Presence Deviation

Unexpected presence/absence: operator leaving mandatory zone, unauthorized entry.

Posture Deviation

Unsafe body positions: bending into machinery, reaching beyond safe zones, lying on ground.

🚨

Alert System

Real-time notifications when detection agents identify conditions exceeding safety thresholds. Classified by type and severity, stored for audit trail.

Alert Types

  • Movement Alert: Operator instability detected over sustained frames
  • Leak Alert: Oil/fluid leakage exceeds area threshold
  • Anomaly Alert: AI Vision identifies unusual conditions

Alert Lifecycle

1Detection
2Creation
3Storage
4Notification
5Acknowledge
6Resolved

Severity Levels

LowMinor deviation, informational MediumNotable deviation, attention recommended HighSignificant concern, intervention required CriticalSevere hazard, immediate action mandatory