Invoice Processing Tool Documentation

Overview

The Invoice Processing Tool is a web-based application designed to automate the extraction of information from invoice images. It supports both typed and handwritten invoices through the use of Tesseract OCR (local) and Google Cloud Vision API (cloud). The extracted data includes invoice number, date, vendor details, customer information, item breakdown, and totals.

The tool consists of:

• Flask API backend for OCR and parsing logic
• React.js frontend for user interaction

---

Table of Contents

1. Features
2. Technologies Used
3. Project Structure
4. Installation
   • Prerequisites
   • Setup
5. Running the Application
6. API Endpoints
7. Example Usage
8. Additional Scripts
9. License
10. Acknowledgements

---

Features

Text Extraction

• Extracts raw text using Tesseract OCR or Google Cloud Vision API

Invoice Parsing

• Extracts and identifies fields:
  • Invoice Number
  • Date
  • Vendor and Customer Names/Addresses
  • Item List
  • Subtotal, Tax, Total

File Upload

• Upload invoices via frontend

CORS Support

• Allows cross-origin requests from the React.js frontend

Error Handling

• Falls back to Tesseract OCR if Google Vision fails

---

Technologies Used

• Python 3.7+ - Backend logic
• Flask - API development
• Tesseract OCR - Local OCR engine
• Google Cloud Vision API - Cloud OCR engine
• React.js - Frontend framework
• Flask-CORS - CORS middleware for Flask
• Werkzeug - File upload handling

---

Project Structure

Invoice_Processing/
├── Backend/             # Flask API backend
│   ├── app.py           # Main entry point for Flask app
│   ├── test_vision.py   # Script to test Vision API
│   ├── test.http        # Optional HTTP request testing file
│   └── train_donut.py   # Document-understanding model trainer
├── Frontend/            # React frontend
│   ├── node_modules/    # React dependencies
│   ├── src/             # Frontend source files
│   └── README.md        # Frontend documentation
├── public/              # Static assets (images, etc.)
├── results/             # Output results and parsed data
├── src/                 # Shared or top-level source files
├── uploads/             # Uploaded invoice images
├── .env/                # Virtual environment (should be gitignored)
├── .gitignore
├── README.md            # Project documentation
├── requirements.txt     # Python dependencies
├── package.json         # Node.js dependencies
├── package-lock.json
├── postcss.config.js
└── .gitattributes

---

Installation

Prerequisites

• Python 3.7 or higher
• Tesseract OCR installed and accessible via PATH
• Google Cloud Platform account (for Vision API)
• Flask and other required Python packages

Setup

1. Clone the repository:

git clone https://github.com/yourusername/invoice-processing-tool.git
cd invoice-processing-tool

2. Install Python dependencies:

pip install -r requirements.txt

3. Install Tesseract OCR:

• Windows: Download from Tesseract GitHub, add to system PATH
• Linux:

sudo apt install tesseract-ocr

4. Setup Google Vision API:
   • Create a GCP project
   • Enable Vision API
   • Download service account key JSON
   • Set the environment variable:

export GOOGLE_APPLICATION_CREDENTIALS="/path/to/credentials.json"

5. Create an uploads directory:

mkdir uploads

---

Running the Application

Start Flask Backend:

cd Backend
python app.py

Runs on: http://localhost:5000

Start React Frontend:

cd Frontend
npm install
npm start

Runs on: http://localhost:3000

---

API Endpoints

GET /

• Renders a basic landing HTML page

POST /process-invoice

• Accepts: Form-data with key invoice (image file)
• Returns: JSON with raw extracted text

Example Request:

invoice=<image-file>

Example Response:

{
  "raw_text": "Extracted text from the invoice"
}

---

Example Usage

1. Upload invoice via frontend form
2. Backend uses OCR (Vision API → Tesseract fallback)
3. Text is parsed and returned as structured JSON

Example Raw Text Output:

Invoice No: INV123456
Date: 2025-04-30
Vendor: Example Vendor Inc.
Customer: John Doe
Items:
- Item 1: $100.00
- Item 2: $50.00
Subtotal: $150.00
Tax: $15.00
Total Amount: $165.00

---

Additional Scripts

• train_donut.py: Used for training a document-understanding model (Donut). Useful if you want to improve accuracy beyond default OCR methods.
• test_vision.py: Quick script to validate Google Vision API results.
• test.http: Optional file for testing HTTP requests to the Flask API.

---

License

This project is licensed under the MIT License. See the LICENSE file for more details.

---

Acknowledgements

• Tesseract OCR - GitHub
• Google Cloud Vision API - Documentation
• Flask - Official Docs
• Werkzeug - Documentation

---