Receipt Analyzer

An intelligent OCR-based receipt analysis tool that extracts key information from receipt images.

Features

• Extracts merchant name, date, amount, and currency
• Supports multiple receipt types (invoice, receipt, order)
• Handles multiple currency formats (USD, EUR, GBP, CNY, HKD, JPY)
• Provides confidence scores for extracted data
• Includes fuzzy matching for better accuracy
• Caches results for improved performance

Requirements

• Python 3.7+
• doctr
• matplotlib
• fuzzywuzzy
• python-Levenshtein (optional, for better performance)

Installation

1. Clone the repository:

git clone https://github.com/xer0bit/receipt-analyzer.git
cd receipt-analyzer
pip install -r requirements.txt

Usage

Basic usage example:

from analyzer import analyze_receipt

result = analyze_receipt('path/to/receipt.jpg')
print(result)

Output Format

The analyzer returns a dictionary containing:

• merchant_name: Extracted business name
• bill_date: Date in YYYY-MM-DD format
• amount: Transaction amount
• currency: Detected currency code
• description: Transaction description
• type: Receipt type (invoice/receipt/order)
• confidence_score: Confidence level (0-1)

Configuration

Customize the behavior by modifying config.py:

• Supported currencies
• Keywords for receipt classification
• Path configurations

Error Handling

The analyzer includes validation checks and will raise ValueError for invalid receipts. Always wrap the analyze_receipt call in a try-except block:

try:
    result = analyze_receipt('path/to/receipt.jpg')
except ValueError as e:
    print(f"Error: {e}")

API Usage

Start the API server:

API Testing with Postman

1. Start the API server: