Skip to content

README.md

Latest commit

 

History

History
288 lines (201 loc) · 5.72 KB

README.md

File metadata and controls

288 lines (201 loc) · 5.72 KB

🧾 Invoice Data Extractor (v2.0.0)

GitHub release (latest by date) MIT License Platform AI Engine

A desktop-based AI invoice processing engine that extracts structured, item-level invoice data from PDF and JPG invoices using a hybrid OCR + LLM pipeline, and exports it directly to Excel.

✔ Tested on 1000+ real-world invoices ✔ Achieved ~91% field-level accuracy ✔ Packaged as a single plug-and-play .exe

🎯 Problem This Solves

Invoice data extraction is messy because:

  • PDFs vary wildly in layout
  • Many PDFs are scanned images
  • Tables break traditional parsers
  • OCR alone is not enough
  • Rule-based systems don’t scale

This tool solves that by combining:

  • Deterministic OCR & preprocessing
  • LLM-based semantic understanding
  • Structured output enforcement
  • Excel-ready data flattening

✨ What’s New in v2.0.0

This release is a major backend upgrade.

🔥 Major Improvements

  • Groq API integration

    • Replaced Gemini with Groq (LLaMA-3.1-8B-Instant)
    • Faster inference, lower latency, better consistency

  • Advanced OCR Pipeline

    • OpenCV preprocessing (denoise, CLAHE, thresholding)
    • High-DPI PDF to image conversion
    • Custom Tesseract OCR configuration

  • Robust PDF Handling

    • pdfplumber text extraction first
    • Automatic OCR fallback if text is missing

  • Flattened Excel Output

    • Each invoice item becomes one row
    • Invoice-level fields repeated per item
    • Suitable for accounting & analytics

  • Append to Existing Excel

    • New invoices can be appended to existing files
    • No overwriting, no manual merges

🧠 System Architecture

PDF / JPG
   │
   ├─► pdfplumber (text PDFs)
   │
   └─► OpenCV + Tesseract (scanned PDFs / images)
            │
            ▼
     Cleaned invoice text
            │
            ▼
     Groq LLM (LLaMA-3.1-8B)
            │
            ▼
     Structured JSON (validated)
            │
            ▼
     Flattened rows
            │
            ▼
     Excel (.xlsx)

🧠 How the Extraction Works (Step-by-Step)

1️⃣ PDF / Image Ingestion

  • PDFs are first parsed using pdfplumber
  • If no usable text is found → OCR pipeline is triggered
  • JPG files always go through OCR

2️⃣ OCR Enhancement (Key to Accuracy)

Each page/image undergoes:

  • Grayscale conversion
  • Noise removal
  • CLAHE contrast enhancement
  • Adaptive thresholding
  • Custom Tesseract config for invoice symbols & numbers

This significantly improves OCR quality on:

  • Scanned invoices
  • Low-contrast documents
  • Poorly printed bills

3️⃣ LLM-Based Data Extraction (Groq)

Extracted text is sent to Groq with a strict, structured prompt that enforces:

  • Exact text extraction (no hallucination)

  • Fixed fields:

    • Company name
    • Invoice number
    • Invoice date
    • FSSAI number
    • Item-level details

  • JSON-only response

  • Clean numeric values

  • Standardized date format

Model used: llama-3.1-8b-instant

4️⃣ Response Cleaning & Validation

  • Markdown fences removed
  • JSON boundaries detected manually
  • Invalid responses discarded
  • Safe fallback handling for malformed outputs

5️⃣ Data Flattening Logic

  • Each invoice item becomes one Excel row
  • Invoice-level fields are repeated per item
  • If no items are found → invoice-level row is still created

This makes the output:

  • Pivot-friendly
  • BI-ready
  • Accounting-friendly

6️⃣ Excel Export (Append-Safe)

  • If Excel file exists → data is appended
  • Column mismatches are auto-handled
  • Weights are converted to kilograms
  • File is saved using openpyxl

📊 Accuracy & Testing

  • ✅ Tested on 1000+ real invoices

  • 📈 Achieved ~91% accuracy

  • Includes:

    • Printed PDFs
    • Scanned PDFs
    • Multi-item invoices
    • Inconsistent layouts

Accuracy measured on:

  • Invoice number
  • Date
  • Item description
  • Quantity
  • Amount
  • HSN codes (when present)

🧪 Supported Input Formats

Format Method
PDF (text) pdfplumber
PDF (scanned) OCR fallback
JPG OCR

📤 Output

Excel File (.xlsx)

Each row represents one invoice item.

Columns include:

  • company_name
  • invoice_number
  • invoice_date
  • fssai_number
  • description
  • hsn_code
  • quantity
  • weight (kg)
  • rate
  • amount

🚀 Getting Started

🔽 Download Executable

👉 Download v2.0.0

No Python. No setup. Just run.

🔑 Groq API Key (Required)

  1. Create a Groq account
  2. Generate an API key
  3. Paste it when prompted by the app

🛠 Usage Flow

  1. Launch the .exe
  2. Enter Groq API key
  3. Select one or multiple PDF/JPG invoices
  4. Choose output Excel file (new or existing)
  5. Extraction starts automatically
  6. Excel file is generated/appended

🧪 Build From Source

git clone https://github.com/Cherry28831/Invoice-Data-Extractor.git
cd Invoice-Data-Extractor
pip install -r requirements.txt
cd backend
pyinstaller invoice-backend.spec
cd ..
npm start 
npm run dist

🌱 Future Roadmap

  • Vendor auto-classification
  • GST / tax breakup extraction
  • CSV & ERP exports
  • Multi-language invoices
  • Cloud + desktop hybrid mode

🤝 Contributions

This project is actively evolving. Contributions, optimizations, and feedback are welcome.