Historical Document Interpreter 📜🔍

A Flask-based web application that digitizes and interprets historical handwritten documents using the Gemini AI API

🌟 Features

📥 Document Upload

• Accepts PDFs, JPGs, and PNGs of historical documents
• Handles multi-page PDFs by converting each page to images
• Secure file storage with original preservation

✨ AI-Powered Processing (Gemini API)

• Handwriting Recognition: Extracts text from even poor-quality scans
• Text Cleaning: Fixes spacing and obvious errors while preserving original meaning
• Smart Entity Extraction: Identifies and tags:
  • People ([PERSON])
  • Locations ([LOCATION])
  • Dates ([DATE])
  • Document types ([DOC_TYPE])
• Automatic Summarization: Creates easy-to-understand summaries
• Historical Term Glossary: Explains archaic legal/administrative terms

💬 Interactive Document Chat

• Ask natural language questions like:
  • "Who is mentioned in this document?"
  • "What is the main purpose of this deed?"
  • "Translate this section to Hindi"
• Get context-aware answers based on the document content

📊 Results Presentation

• Side-by-Side View: Compare original document with processed text
• Color-Coded Highlights: Easily spot important entities
• Download Options: Save processed text or summary as TXT/PDF

⏳ Processing History

• View all previously processed documents
• Quick access to past results
• Never lose your work

🏆 Use Cases

🏛️ For Historians & Researchers

• Quickly digitize archival materials
• Search across collections for names/places
• Understand obsolete legal terminology

👨‍⚖️ Legal Professionals

• Analyze old property deeds and land records
• Trace historical ownership chains
• Extract key clauses from vintage contracts

👪 Genealogy Enthusiasts

• Read old family documents and letters
• Identify ancestors in historical records
• Understand context of old certificates

📚 Librarians & Archivists

• Create searchable digital archives
• Preserve fragile documents through digitization
• Make collections more accessible

🛠️ Technical Stack

Backend

• Flask (Python web framework)
• SQLite (Lightweight database)
• Gemini API (For all AI processing)
• PyMuPDF (PDF processing)
• Pillow (Image handling)

Frontend

• HTML5/CSS3 with responsive design
• JavaScript for interactive features
• Bootstrap 5 for clean UI components

🚀 Getting Started

Prerequisites

• Python 3.8+
• Gemini API key (free tier available)
• Basic terminal skills

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/historical-docs-interpreter.git
   cd historical-docs-interpreter

2. Install dependencies:

   pip install -r requirements.txt

3. Configure your Gemini API key:

   • Create a config.py file
   • Add: GEMINI_API_KEY = "your-api-key-here"

4. Run the application:

   python app.py

5. Access in your browser at:

   http://localhost:5000
   

📖 Usage Guide

1. Upload Documents

   • Click "Choose File" or drag-and-drop
   • Supported formats: PDF, JPG, PNG
   • Maximum size: 16MB

2. View Results

   • Left panel shows original document
   • Right panel shows processed text with:
     • Summary at the top
     • Full text below
     • Color-highlighted entities

3. Ask Questions

   • Type questions in the chat box
   • Examples:
     • "List all people mentioned"
     • "What year was this written?"
     • "Explain the legal terms used"

4. Access History

   • View all processed documents
   • Click any entry to reload results

🌍 Multilingual Support

The system can handle documents in:

• English (including archaic forms)
• Hindi (and other Indian languages)
• Mixed-language documents

Translation features:

• Modernize old language forms
• Explain regional terms
• Cross-language queries ("Translate this to Hindi")

⚠️ Limitations

1. Handwriting Quality

   • Works best with clear handwriting
   • May struggle with very cursive scripts

2. Document Age

   • Best for 18th-20th century documents
   • Older materials may need manual verification

3. Language Coverage

   • Strongest support for English and Hindi
   • Other languages may have reduced accuracy

📜 License

MIT License - Free for personal and academic use

🙏 Acknowledgments

• Built using Google's Gemini AI
• Inspired by archival digitization projects worldwide
• Community contributors welcome!

---