🏗️ Drilling Site Safety Detection System

Video, PPE and heatmaps Violation log

🎯 Model Configuration

📦 Model 1: General Objects (Person, Vehicles)

✅ Current: C:\Users\Виктор\source\repos\Crane_detection\Crane_detection\runs\detect\crane_core_custom\yolo11l.pt

🏗️ Model 2: Safety Equipment (Hook, Load, PPE, Harness)

✅ Current: C:\Users\Виктор\source\repos\Crane_detection\Crane_detection\runs\detect\crane_enhanced_v2_21_11\weights\best.pt
✓ Model 1 loaded | Model 2 loaded

🎥 Video Source

📁 Local File
🌐 URL
📤 Upload
📡 RTSP
📹 MP4 File: 20251217_123630_D01_20251023111831.mp4

Frame 916 / 7314 (12.5%)

🎯 Detection Results

📏 Current Scale: 1 pixel = 0.0200 meters
Mode: AUTO
Camera Angle: 10.0°
Horizon Position: 59% from top
10m = 500 pixels
📐 Perspective Settings:
Camera Angle: 10.0° (0°=horizontal, 90°=top-down)
Horizon Line: 59% from top
Ground surface: From horizon to bottom of image

Waiting for a frame…

Processing: 164ms · Size: 2304x1296

📍 Isometric Safety Map (Top View)

Waiting for the map…

⚠️ Safety Alerts

✓ No safety violations detected
General Objects (Model 1)1
Safety Equipment (Model 2)1
Danger Zones1
Total Alerts0
Critical Alerts0
High Priority0

📐 Perspective & Calibration

📷 Camera Perspective

0° = horizontal view, 90° = top-down view
Position from top (1% to 99%). Ground surface extends from horizon to bottom.

📏 Distance Calibration

⚠️ Danger Zone Rules

🏗️ Crane Zone

🪝 Hook Projection Zone

📦 Load Zone

🚗 Vehicle Proximity

🧗 Height Safety

🔔 Alarm System

🔕 Alarm System Disabled
Leave empty if Arduino doesn't require token
Protocol Guide:
• Arduino TCP: Sends "ALARM <ms> [token] " (compatible with activator_reley.py)
• HTTP: REST API POST request
• TCP JSON: JSON message over TCP
• Modbus TCP: Industrial protocol (requires pymodbus)

🎨 Image Enhancement

⚙️ Processing Settings

Images larger than this will be scaled down for processing Objects smaller than this area will be filtered out Non-Maximum Suppression overlap threshold