Invoice OCR - Google Vision API Boilerplate

Overview

This project provides a boilerplate codebase for extracting invoice details using Google Vision OCR API. It processes invoices (PDFs or images), extracts structured data (invoice number, total amount, GST, etc.), and stores it in a MySQL database.

This template allows users to quickly integrate Google OCR by adding their API key and installing dependencies.

---

Features

• Google Vision OCR Integration: Uses Google Cloud’s OCR for extracting text from invoices.
• Image & PDF Support: Converts PDFs to images and processes them.
• Preprocessing Enhancements: Includes adaptive thresholding, contrast improvement, and deskewing.
• Structured Data Extraction: Extracts invoice number, date, total amount, and GST details.
• MySQL Database Integration: Stores extracted invoice data in a structured format.
• Frontend Integration: Simple HTML form for uploading invoices.

---

Prerequisites

1. Python 3.8+
2. Google Cloud API Key (for Vision OCR)
3. MySQL Database (if storing extracted data)
4. Flask for Backend
5. Basic HTML & JavaScript for Frontend

---

Setup Instructions

1️⃣ Clone the Repository

git clone https://github.com/your-username/Google-Vision-OCR.git
cd Google-Vision-OCR

2️⃣ Install Dependencies

Install the required Python packages:

pip install -r backend/requirements.txt

If requirements.txt is missing, manually install:

pip install flask flask-cors mysql-connector-python pillow pytesseract pdf2image opencv-python numpy requests python-dotenv

3️⃣ Set Up Your Google Cloud API Key

1. Obtain your API key from Google Cloud Console.
2. Create a .env file inside the backend/ folder:

   API_KEY=your-google-cloud-api-key
   

3. The backend will automatically load the API key from .env.

4️⃣ Start the Backend

cd backend
python app.py

This will start the Flask API at http://127.0.0.1:5000.

5️⃣ Start the Frontend

Simply open frontend/index.html in a browser.

---

API Endpoints

1️⃣ Upload Invoice Image

• Endpoint: POST /upload-invoice
• Description: Accepts an image or PDF, extracts text using Google Vision OCR, and returns structured invoice data.
• Request Format:
  • Form-data:
    • file: Invoice image or PDF file.
• Response Example:

  {
    "status": "success",
    "extracted_data": {
      "invoice_number": "INV-12345",
      "invoice_date": "2024-01-10",
      "invoice_amount": "3450.00",
      "gst_no": "24ABCDE1234F1ZP",
      "gst_percentage": "5"
    }
  }

2️⃣ Save Extracted Invoice Data

• Endpoint: POST /confirm-invoice
• Description: Saves extracted invoice details into MySQL.
• Request Format (JSON):

  {
    "invoice_number": "INV-12345",
    "invoice_date": "2024-01-10",
    "invoice_amount": "3450.00",
    "gst_no": "24ABCDE1234F1ZP",
    "gst_percentage": "5"
  }

• Response:

  { "status": "success", "message": "Invoice data saved successfully" }

---

Project Structure

/Google-Vision-OCR
│── backend/
│   ├── app.py               # Main Flask API
│   ├── ocr_processor.py     # Image preprocessing & text extraction
│   ├── utils.py             # Helper functions
│   ├── requirements.txt     # Python dependencies
│   ├── .env                 # API key (not committed)
│   ├── static/              # Static assets
│   ├── templates/           # Flask templates (if needed)
│   ├── uploads/             # Temporary storage for uploaded invoices
│── frontend/
│   ├── index.html           # Basic UI for uploading invoices
│   ├── script.js            # Handles API calls to backend
│   ├── styles.css           # Basic styling
│── .gitignore               # Ensures API key & sensitive files aren't committed
│── README.md                # Project documentation
│── LICENSE                  # MIT License (allows open-source usage)

---

Deployment Instructions

If deploying on Heroku, Vercel, or AWS, set up environment variables instead of a .env file.

Example for Heroku:

heroku config:set API_KEY=your-google-cloud-api-key

Then start the Flask server:

gunicorn backend.app:app

---

Security Considerations

• Never expose your API key in frontend code or public repositories.
• Use .gitignore to prevent committing .env:

  .env
  

• Rotate API keys periodically to prevent misuse.

---

License

This project is licensed under the MIT License, allowing free use, modification, and distribution with attribution.

---

Contributing

Feel free to contribute by:

• Improving OCR accuracy (e.g., better regex patterns).
• Adding support for more invoice formats.
• Enhancing the frontend UI.

---

Acknowledgments

This project is based on Google Vision API and various open-source tools like Tesseract OCR, Flask, and OpenCV.

---