RedactAI: Advanced Document Redaction System

RedactAI is a powerful document redaction system designed to protect sensitive information in various types of documents. Using advanced OCR technology and AI-assisted detection, RedactAI makes it easy to identify and redact personal information from PDF files and images.

🌟 Unique Value Proposition

• Smart Document Analysis: Automatically detects sensitive information using AI-powered analysis
• Multiple Redaction Methods: Choose between AI-assisted redaction or text-based redaction
• Customizable Redaction Types: Apply temporary or permanent redactions based on your needs
• User-friendly Interface: Intuitive web interface for easy document processing

📋 Features

• Document Upload: Support for PDF and image formats (JPG, PNG, TIFF, etc.)
• OCR Technology: Extract text from scanned documents and images
• AI Analysis: Automatically identify sensitive information like names, addresses, ID numbers
• Field Selection: Select specific fields detected by AI for redaction
• Text-based Redaction: Type specific text to find and redact across the document
• Permanent/Temporary Redaction: Choose between black boxes (permanent) or yellow highlights (temporary)
• Multi-language Support: Process documents in multiple Indian languages
• Downloadable Results: Download the redacted document in its original format

Project Structure

RedactAI/
├── ocr/
│   ├── backend/
│   │   ├── app.py                # Flask backend server
│   │   └── requirements.txt      # Python dependencies
│   └── frontend/
│       ├── public/               # Static files
│       └── src/                  # React source code
│           ├── components/       # React components
│           │   ├── Header.js     # App header
│           │   ├── UploadPage.js # File upload page
│           │   └── EditorPage.js # Document editor page
│           ├── App.js            # Main App component
│           └── index.js          # Entry point

🚀 Getting Started

Prerequisites

• Python 3.8+
• Node.js 14+
• Tesseract OCR with language data files

Backend Setup

# Clone the repository
git clone https://github.com/Boobeshkumar56/RedactAI.git
cd RedactAI/ocr

# Create and activate virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install Python dependencies
pip install -r requirements.txt

# Run the backend server
cd backend
python app.py

Frontend Setup

# Navigate to frontend directory
cd ../frontend

# Install dependencies
npm install

# Run the development server
npm start

The application should now be accessible at http://localhost:3000

📚 API Documentation

RedactAI offers a comprehensive API for document redaction:

1. Upload Document

Endpoint: POST /api/upload

Content-Type: multipart/form-data

Request Parameters:

• file: The document file (PDF, JPG, PNG, etc.)
• language: Language code ('eng', 'tam', 'hin', 'tel', 'kan', 'mal', or combinations like 'eng+tam')

Response:

{
  "file_id": "unique-file-identifier.pdf",
  "original_filename": "original-name.pdf",
  "language": "eng",
  "data_fields": [
    {
      "id": "field-id",
      "text": "Extracted text",
      "page": 0,
      "confidence": 90,
      "position": {
        "x": 100,
        "y": 200,
        "width": 300,
        "height": 50
      }
    }
  ]
}

2. Apply Redactions

Endpoint: POST /api/redact

Content-Type: application/json

Request Parameters:

For field selection redaction:

{
  "file_id": "unique-file-identifier.pdf",
  "redactions": [
    {
      "id": "field-id",
      "text": "Text to redact",
      "page": 0,
      "redaction_type": "temporary",
      "position": {
        "x": 100,
        "y": 200,
        "width": 300,
        "height": 50
      }
    }
  ],
  "redaction_type": "temporary",
  "language": "eng"
}

For text-based redaction:

{
  "file_id": "unique-file-identifier.pdf",
  "text_to_redact": [
    {
      "text": "John Doe",
      "redaction_type": "permanent",
      "case_sensitive": false
    }
  ],
  "document_type": "unknown",
  "language": "eng"
}

Response:

{
  "redacted_file_id": "redacted_unique-file-identifier.pdf",
  "total_redactions": 3,
  "redaction_method": "manual_typing"
}

3. Download Redacted Document

Endpoint: GET /api/download/{file_id}

Response: The redacted document file.

4. AI Document Analysis

Endpoint: POST /api/analyze-document

Content-Type: application/json

Request Parameters:

{
  "file_id": "unique-file-identifier.pdf",
  "language": "eng"
}

Response:

{
  "sensitive_fields": [
    {
      "id": "ai-field-123",
      "text": "John Doe",
      "category": "Name",
      "ai_confidence": 95,
      "position": {
        "x": 100,
        "y": 200,
        "width": 300,
        "height": 50
      }
    }
  ],
  "analysis_type": "gemini_ai"
}

🔮 Future Enhancements

1. Document Type Selection: Add support for selecting document types to enable template-based redaction for standard documents like Aadhaar cards, PAN cards, and passports.

2. Manual Sketching Tool: Implementing a drawing-based redaction tool to allow users to manually mark areas for redaction using a pen or brush tool.

3. Custom Template Creation: Allow users to create and save their own templates for repeated document types.

4. Batch Processing: Support for uploading and processing multiple documents at once.

5. Document Classification AI: Automatically detect document type without user input.

6. Enhanced Security Features: Add watermarking, encryption, and audit logging for redacted documents.

7. Mobile Application: Develop mobile apps for on-the-go document redaction.

8. Integration with Cloud Storage: Direct integration with Google Drive, Dropbox, etc.

9. Advanced OCR Enhancement: Implement pre-processing techniques to improve OCR accuracy for low-quality scans.

10. Browser Extension: Create a browser extension for quick redaction of online documents.

11. Enterprise Features: Role-based access control, organization management, and compliance reporting.

🔄 Workflow

1. Upload your document (PDF or image)
2. Choose language for OCR processing
3. Select redaction method:
   • AI Analysis: Automatically detect and select sensitive information
   • Text-based Redaction: Type specific text to find and redact
4. Choose redaction type (temporary or permanent)
5. Apply redactions
6. Download the redacted document

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

👥 Contributors

• Boobeshkumar56 - Creator and maintainer

Images: