extract_plate_from_image

​

Overview

​

Function Signature

def extract_plate_from_image(image_path: str) -> str

​

Parameters

​

Return Value

​

Implementation Details

​

Image Processing Pipeline

1. Base64 Encoding: The image file is read in binary mode and encoded to base64 format for transmission to the OpenAI API.
2. GPT-4o Vision Analysis: The encoded image is sent to the gpt-4o-mini model with a specialized prompt in Spanish that instructs the model to extract only the license plate text without additional commentary.
3. Text Sanitization: The response is processed through multiple cleaning steps:
   • Stripped of whitespace
   • Converted to uppercase
   • Filtered to keep only alphanumeric characters and hyphens
   • Truncated to maximum 10 characters
4. Error Handling: Any exceptions during the OCR process are caught, logged to console, and result in returning the fallback value "NO_DETECTADA".

​

OpenAI API Configuration

model="gpt-4o-mini"
messages=[
    {
        "role": "user",
        "content": [
            {
                "type": "text",
                "text": "Extrae únicamente el texto de la matrícula visible..."
            },
            {
                "type": "image_url",
                "image_url": {"url": "data:image/jpeg;base64,..."}
            }
        ]
    }
]

​

Code Example

from app import extract_plate_from_image

# Extract plate from a saved image
image_path = "/path/to/uploads/matricula_20240315_143022.jpg"
plate_number = extract_plate_from_image(image_path)

if plate_number != "NO_DETECTADA":
    print(f"License plate detected: {plate_number}")
    # Save to database or process further
else:
    print("No license plate could be detected")
    # Fallback to alternative OCR method

​

Usage in Application

@app.route("/guardar", methods=["POST"])
def guardar():
    # ... file upload handling ...
    
    # Primary OCR with GPT-4o
    matricula = extract_plate_from_image(path)
    
    # Fallback to pytesseract if GPT-4o fails
    if matricula == "NO_DETECTADA":
        image = Image.open(path)
        ocr_text = pytesseract.image_to_string(image, lang="eng")
        ocr_text = ocr_text.strip().replace(" ", "").replace("\n", "").upper()
        matricula = "".join([c for c in ocr_text if c.isalnum() or c == "-"])[:10]
    
    # ... save to CSV ...

​

Error Handling

​

Environment Requirements

​

Dependencies

from openai import OpenAI
import base64
import os
from dotenv import load_dotenv

​

Configuration

load_dotenv()
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

OPENAI_API_KEY=sk-your-api-key-here

​

Performance Considerations

• API Latency: Typical response time is 2-5 seconds depending on image size and API load
• Cost: Each call consumes OpenAI API credits (gpt-4o-mini pricing applies)
• Image Size: Larger images take longer to encode and transmit; consider resizing images before processing
• Rate Limits: OpenAI enforces rate limits; implement request queuing for high-volume scenarios

​

Best Practices

1. Image Quality: Ensure uploaded images are clear and well-lit for optimal extraction accuracy
2. Error Recovery: Always implement fallback OCR methods (e.g., pytesseract) for robustness
3. Logging: Monitor console output for OCR errors to track API issues
4. Validation: Validate extracted plate numbers against expected formats for your region
5. Caching: Consider caching results to avoid reprocessing the same images

​

Related Functions

• read_csv() - Read stored license plate records
• write_csv() - Save extracted plates to persistent storage
• ensure_csv() - Initialize CSV storage structure