Asma-da/OCR-Tool-for-Tunisian-Documents-

# Tunisian Document OCR & Verification Platform A secure, fully on-premise Optical Character Recognition (OCR) system designed specifically for Tunisian official documents. The platform extracts text, verifies document authenticity, and exports structured data while ensuring data privacy and regulatory compliance. --- ## 📌 Project Overview This project automates the processing of Tunisian administrative documents by combining OCR, data validation, and authenticity verification in a single platform. It supports bilingual content (Arabic & French) and addresses the unique challenges of Tunisian documents such as cursive Arabic script, security elements, and varying document quality. --- ## ✨ Key Features - **🏠 Fully On-Premise Deployment** All processing is done locally , no cloud dependency for enhanced data privacy. - **🧠 Advanced OCR Extraction** Accurate text extraction from images and PDFs with Arabic and French support. - **🔍 Document Authenticity Verification** - Format validation (CIN, Passport numbers) - Logical consistency checks (e.g., date of issue > date of birth) - Completeness and integrity validation - Confidence scoring for reliability assessment - **🔐 Secure Access Control** - JWT-based authentication - Role-based access control (Admin / User) - **📤 Multi-Format Data Export** Export extracted data and verification results to: PDF, Excel, CSV, JSON - **🌐 RESTful API** Easily integrates with existing systems and workflows. --- ## 📄 Supported Documents - 🇹🇳 Tunisian National Identity Card (CIN) – Front & Back - 🛂 Tunisian Passport - 📑 PDF Documents (e.g., contracts, administrative files) --- ## 🧩 Tunisian Document Specificities - Bilingual content - Right-to-left cursive Arabic script - Security elements (watermarks, stamps, background patterns) - Machine Readable Zone (MRZ) for passports - Variable quality due to scans and mobile captures --- ## 🛠️ Tech Stack **Backend** - Python (FastAPI / Flask) - OCR Engines (Tesseract / custom …