Traffic Violation Detection System

An intelligent traffic monitoring system that detects multiple types of traffic violations, tracks vehicles, and manages traffic signals based on traffic density.

Features

• Vehicle Detection and Tracking

  • Real-time detection of cars, motorcycles, buses, and trucks
  • Vehicle tracking with unique IDs
  • Speed calculation for each vehicle

• Violation Detection

  • Speeding violations
  • Lane crossing detection
  • Helmet usage monitoring for motorcycles
  • Triple riding detection on motorcycles
  • License plate recognition and tracking

• Smart Traffic Signal System

  • Adaptive traffic signal timing based on vehicle density
  • Different weightage for various vehicle types:
    • Cars: 1.0
    • Motorcycles: 0.5
    • Buses: 2.0
    • Trucks: 1.5
  • Real-time vehicle counting when crossing detection line
  • Visual traffic signal display with timing information

• Criminal Vehicle Detection

  • License plate matching against database of criminal vehicles
  • Real-time alerts for detected criminal vehicles
  • MongoDB integration for criminal records

Prerequisites

# Required Python packages
pip install torch torchvision
pip install ultralytics
pip install easyocr
pip install opencv-python
pip install pymongo
pip install numpy
pip install sort-tracker

Required Model Files

Place these model files in the same directory as the script:

• yolov8n.pt - General object detection
• helmet_detection.pt - Helmet detection
• license_plate_detector.pt - License plate detection

Configuration

Key parameters that can be adjusted in the code:

# Speed and timing constants
FPS = 30
SPEED_LIMIT = 60  # km/h
MIN_GREEN_TIME = 20  # seconds
MAX_GREEN_TIME = 60  # seconds
YELLOW_TIME = 3  # seconds
RED_TIME = 5  # seconds

# Vehicle detection settings
MIN_CONFIDENCE = 0.5
VEHICLE_THRESHOLD = 5  # Minimum vehicles for traffic density calculation

Usage

python final1.py [options]

Options:
  --video PATH           Path to video file or 0 for webcam (default: 'sample.mp4')
  --criminal_plates PATH Path to CSV file containing criminal license plates
  --output_csv PATH     Path for saving detected license plates

Output

1. Visual Display

   • Bounding boxes around detected vehicles
   • Vehicle IDs and types
   • Speed measurements
   • Violation warnings
   • Traffic signal status
   • Vehicle counts and weighted traffic density

2. CSV Output

   • Timestamp
   • License plate number
   • Confidence score
   • Vehicle type
   • Speed
   • Detected violations

Database Integration

The system uses MongoDB to store and check criminal vehicle records:

• Database: 'vehicle_data'
• Collection: 'criminal_records'
• Connection: localhost:27017

File Structure

traffic_system/
├── FINAL/
│   ├── final1.py          # Main script
│   ├── sort.py            # SORT tracking implementation
│   └── criminal_plates.csv # List of criminal plates
├── model/
│   ├── yolov8n.pt
│   ├── helmet_detection.pt
│   └── license_plate_detector.pt
└── README.md

Violation Detection Logic

1. Speed Violation

   • Calculated using pixel-to-meter conversion
   • Triggers when speed > SPEED_LIMIT

2. Lane Crossing

   • Detected when vehicle crosses boundary line
   • Uses center point of vehicle bounding box

3. Helmet Detection

   • Applied to detected motorcycles
   • Uses dedicated helmet detection model

4. Triple Riding

   • Counts persons on motorcycles
   • Triggers when count > 2

Traffic Signal Algorithm

1. Signal changes based on:

   • Vehicle count
   • Weighted vehicle density
   • Minimum and maximum timing constraints

2. Timing calculation:

   if weighted_count > VEHICLE_THRESHOLD:
       green_time = min(MAX_GREEN_TIME, MIN_GREEN_TIME + (weighted_count - VEHICLE_THRESHOLD) * 2)
   else:
       green_time = MIN_GREEN_TIME

Known Limitations

• Requires good lighting conditions for optimal detection
• License plate recognition accuracy depends on image quality
• Vehicle speed calculation assumes fixed camera position
• Requires calibration for different camera angles

Contributing

Feel free to submit issues, fork the repository, and create pull requests for any improvements.

License

This project is licensed under the MIT License - see the LICENSE file for details.