Full-Field France Vehicle Registration (Carte Grise) OCR

The Challenge: Cryptic EU Codes and Fold Creases

The French vehicle registration certificate (Carte Grise / Certificat d'immatriculation) presents formidable OCR challenges. Like the German Fahrzeugschein, it is a multi-panel folded document, leading to physical creases that distort text in uploaded photos. Furthermore, it relies entirely on EU standard alphanumeric codes instead of explicit labels—for example, 'J.1' designates the vehicle class, 'P.6' is the national tax horsepower, and 'V.9' denotes the environmental class. Standard OCR engines extract these as random strings with no semantic context. Learn more about our robust semantic parsing capabilities on our Vehicle Registration OCR product page.

Semantic Parsing & Dynamic Fallback Delivery

StructOCR is specifically engineered to conquer the complexities of EU documents like the Carte Grise. Our engine uses advanced semantic Key-Value mapping to intelligently locate universal fields (VIN, Make) while effortlessly mapping cryptic EU fields. It seamlessly ignores dense security watermarks and reconstructs text across fold lines. Through our generic fallback architecture, these specialized codes are dynamically captured into a `localized.generic.unmapped_fields` object. Discover how this fits into broader cross-border workflows in our Vehicle Registration Solutions guide.

Core Industry Applications

Integration & API Documentation

Easily integrate our full-field OCR API. For full technical details, visit our Developer API Documentation.

import requests
import base64

# 📸 Test for free with your own image upload: https://structocr.com/developers/vehicle-registration-ocr

with open("carte_grise.jpg", "rb") as image_file:
    base64_image = base64.b64encode(image_file.read()).decode('utf-8')

url = "https://api.structocr.com/v1/vehicle-registration"
headers = {
    "x-api-key": "YOUR_API_KEY",
    "Content-Type": "application/json"
}
payload = {
    "img": base64_image
}

response = requests.post(url, headers=headers, json=payload)
result = response.json()

if result.get('success'):
    data = result.get('data', {})
    std = data.get('standardized', {})
    print("✅ Standardized VIN:", std.get('vin'))
    print("✅ Standardized Plate:", std.get('plate', {}).get('formatted'))
    
    # EU specific codes are captured in the generic unmapped fields fallback
    loc = data.get('localized', {})
    if 'generic' in loc and loc['generic'].get('unmapped_fields'):
        unmapped = loc['generic']['unmapped_fields']
        print("✅ Vehicle Class (J.1):", unmapped.get('j_1'))
        print("✅ Tax Horsepower (P.6):", unmapped.get('p_6'))
        print("✅ Environmental Class (V.9):", unmapped.get('v_9'))

{
  "success": true,
  "data": {
    "document": {
      "country_code": "FR",
      "document_type": "VEHICLE_REGISTRATION",
      "confidence_score": 0.98
    },
    "standardized": {
      "vin": "VF123456789012345",
      "plate": {
        "number": "AA-123-AA",
        "formatted": "AA 123 AA"
      },
      "make_normalized": "Renault",
      "year": 2020
    },
    "localized": {
      "generic": {
        "unmapped_fields": {
          "j_1": "VP",
          "p_6": "5",
          "v_9": "715/2007*2018/1832DG"
        }
      }
    }
  }
}