📇 Business Card Scanner – Automatic Contact Extraction using OCR & NER

🧠 Overview

This project presents an automatic business card scanning system that simplifies the process of adding new contacts from physical cards.
The system extracts and classifies text information from business card images using a hybrid OCR + NER pipeline.

---

🚀 How It Works

1. Image Input
   The user provides an image of a business card.

2. Text Extraction (OCR Stage)
   The image is passed to a pre-trained PaddleOCR model, which extracts all visible text from the card with high accuracy.

3. Entity Recognition (NER Stage)
   The extracted text is then processed by a Named Entity Recognition (NER) model to identify and classify key information such as:

   • 👤 Name / Surname
   • 🏢 Company / Job Title
   • 📧 Email
   • 📞 Phone Number
   • 📍 Address

4. Hybrid Classification Approach
   The NER stage uses a hybrid method that combines:

   • Regular expressions (RegEx) for pattern-based entity detection (emails, phone numbers, etc.).
   • Gliner model — a transformer-based NER architecture that allows custom label definition without retraining, providing flexibility and adaptability across languages and formats.

---

🧩 Technologies Used

Component	Technology
OCR Engine	PaddleOCR
NER Model	Gliner
Text Preprocessing	Python RegEx
Frameworks	Python, PyTorch
Output Format	JSON (structured contact information)

---