ANPR (Automatic Number Plate Recognition) with Database Integration

A comprehensive Python-based license plate recognition system that uses OpenAI's GPT-4o-mini for image analysis and includes Norwegian vehicle registry lookup with SQLite database integration.

🚀 Features

• AI-Powered OCR: Uses OpenAI GPT-4o-mini for accurate license plate text extraction
• Norwegian Registry Integration: Real-time lookup against Norwegian vehicle registry
• Database Tracking: Automatic SQLite database for all registry checks with timestamps
• Smart Upsert: Automatically insert new plates or update existing records
• Comprehensive CLI: Full command-line interface for database management
• Multiple Image Formats: Support for JPEG, PNG, and WebP images
• Robust Error Handling: Comprehensive error handling and logging

📋 Requirements

• Python 3.7+
• OpenAI API key
• Internet connection (for registry lookups)

🛠 Installation

1. Clone the repository:

git clone https://github.com/OsloVision/anpr-pp.git
cd anpr-pp

2. Install dependencies:

pip install -r requirements.txt

3. Set your OpenAI API key:

export OPENAI_API_KEY="your-api-key-here"

📖 Usage

Basic License Plate Reading

Read license plate text from an image:

python license_plate_reader.py path/to/image.jpg

Output:

ABC123

With Norwegian Registry Check + Database Integration

Check license plate against Norwegian registry and automatically save to database:

python license_plate_reader.py --check-registry path/to/image.jpg

Output:

EF49617
hitting url: https://rettsstiftelser.brreg.no/nb/oppslag/motorvogn/EF49617
Pattern found in response.
Norwegian registry: yes
📝 Updated database: EF49617 -> yes

Advanced Options

# Custom API key
python license_plate_reader.py --api-key your-key image.jpg

# Custom database path
python license_plate_reader.py --check-registry --db-path custom.db image.jpg

# Full help
python license_plate_reader.py --help

🗄 Database Management

The system automatically creates a SQLite database to track all registry lookups.

Database Schema

CREATE TABLE loan_status (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    numberplate TEXT NOT NULL,
    loan TEXT NOT NULL,                    -- "yes" or "no" for registry status
    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
    updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
);

Command Line Database Operations

# View all records
python loan_db_utils.py list

# Get specific license plate
python loan_db_utils.py get "EF49617"

# Add manual record
python loan_db_utils.py add "ABC123" "yes"

# Update existing record
python loan_db_utils.py update "ABC123" "no"

# Delete record
python loan_db_utils.py delete "ABC123"

# Check if plate exists in database
python loan_db_utils.py exists "ABC123"

# Create database manually (auto-created when needed)
python loan_db_utils.py create

Programmatic Database Usage

from loan_db_utils import LoanStatusDB

# Initialize database connection
db = LoanStatusDB()

# Add or update records
db.add_loan_status("EF49617", "yes")
db.update_loan_status("EF49617", "no")

# Query records
record = db.get_loan_status("EF49617")
all_records = db.list_all_records()

# Check existence
exists = db.check_numberplate_exists("EF49617")

🔍 Norwegian Registry Integration

The system can check license plates against the Norwegian vehicle registry:

• Registry URL: https://rettsstiftelser.brreg.no/nb/oppslag/motorvogn/{numberplate}
• Pattern Matching: Uses regex to extract registration count
• Results:
  • "yes" - Plate is registered (count > 0)
  • "no" - Plate not found or not registered
• Database Storage: All lookups are automatically saved with timestamps

Registry Check Only

For standalone registry checking without image processing:

from license_plate_reader import check_plate_only

result = check_plate_only("EF49617")  # Returns "yes" or "no"

📁 File Structure

anpr-pp/
├── license_plate_reader.py    # Main ANPR script
├── loan_db_utils.py          # Database utility class and CLI
├── create_database.py        # Database creation script
├── test_regex.py             # Norwegian registry regex tests
├── test_integration.py       # Integration tests
├── test_workflow.py          # Workflow simulation tests
├── requirements.txt          # Python dependencies
├── loan_status.db           # SQLite database (auto-created)
├── DATABASE_INTEGRATION.md  # Detailed integration docs
└── README.md                # This file

🧪 Testing

The project includes comprehensive tests:

# Test Norwegian registry regex
python test_regex.py

# Test database integration
python test_integration.py

# Test full workflow simulation
python test_workflow.py

📷 Supported Image Formats

• JPEG (.jpg, .jpeg)
• PNG (.png)
• WebP (.webp)

📊 Example Workflows

1. Process Multiple Images

#!/bin/bash
for image in *.jpg; do
    echo "Processing $image..."
    python license_plate_reader.py --check-registry "$image"
done

2. Batch Processing with Results

#!/bin/bash
for image in *.jpg; do
    PLATE=$(python license_plate_reader.py "$image")
    echo "$image: $PLATE"
done

3. Database Analysis

from loan_db_utils import LoanStatusDB

db = LoanStatusDB()
records = db.list_all_records()

# Count registered vs unregistered
registered = len([r for r in records if r[2] == "yes"])
unregistered = len([r for r in records if r[2] == "no"])

print(f"Registered: {registered}, Unregistered: {unregistered}")

⚡ API Rate Limits

• OpenAI API: Respects OpenAI's rate limits
• Norwegian Registry: Includes appropriate delays and User-Agent headers
• Timeouts: 10-second timeout for registry requests

🔧 Configuration

Environment Variables

export OPENAI_API_KEY="your-openai-api-key"

Default Settings

• Database Path: loan_status.db (in current directory)
• Request Timeout: 10 seconds
• Image Processing: Temperature 0 for consistent results
• Max Tokens: 50 (sufficient for license plates)

🐛 Error Handling

The system includes comprehensive error handling:

• Missing Files: Clear error messages for missing images
• API Errors: Graceful handling of OpenAI API issues
• Network Issues: Timeout handling for registry requests
• Database Errors: SQLite error handling with rollback
• Image Format: Automatic format detection and fallback

🔒 Privacy & Security

• No Image Storage: Images are processed in memory only
• API Key Security: Support for environment variables
• Database: Local SQLite storage, no external data transmission
• Registry Lookups: Anonymous requests to public registry

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Add tests for new functionality
4. Ensure all tests pass
5. Submit a pull request

📝 License

This project is licensed under the MIT License.

🆘 Support

For issues and questions:

1. Check the Issues page
2. Review the test files for usage examples
3. Check DATABASE_INTEGRATION.md for detailed integration docs

🏗 Architecture

Image Input → OpenAI GPT-4o-mini → License Plate Text
                                        ↓
Norwegian Registry Check ← Regex Pattern Matching
                                        ↓
Database Upsert ← Timestamp Tracking ← Result Processing

📈 Roadmap

• Support for additional vehicle registries
• Batch processing improvements
• Web interface
• Docker containerization
• API endpoint wrapper
• Enhanced image preprocessing