Image to Excel Pipeline (OCR + Vision AI)

A powerful Python pipeline that converts images (invoices, bills, tables) into structured Excel files. It offers two distinct methods: a high-precision Vision AI mode (using Llama 3 via Groq) and a fast, offline Legacy OCR mode (using EasyOCR).

Features

• Vision AI Mode (Recommended):

  • Uses Llama 3 Vision (via Groq API) for superior accuracy.
  • Handles complex layouts, handwritten text, and multi-column tables.
  • Performs logical audits (math validation) on extracted data.
  • Outputs to output/Extracted_Data_Vision.xlsx.

• Legacy OCR Mode:

  • Uses EasyOCR for completely offline processing.
  • Includes spatial analysis for table structure reconstruction.
  • Best for simple, high-contrast documents.
  • Outputs to output/Extracted_Data_OCR.xlsx.

• Excel Formatting:

  • Auto-sized columns.
  • Header styling and color-coding.
  • Validation checks for calculated totals.

Tech Stack

• Language: Python 3.10+
• AI/ML:
  • Groq API (Llama 3 Vision)
  • EasyOCR (PyTorch)
  • OpenCV (Image Preprocessing)
• Data Processing: Pandas, NumPy
• Excel Generation: OpenPyXL

Setup Guide

1. Clone the Repository

git clone <your-repo-url>
cd ImageToExcel

2. Create a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

(Note: If requirements.txt is missing, install manually: pip install groq pandas openpyxl easyocr opencv-python-headless)

4. Configure API Key

Copy the example environment file and add your Groq API key:

cp .env.example .env

Open .env and paste your key:

GROQ_API_KEY=gsk_...

You can get a free key from console.groq.com.

Usage

Option 1: Vision AI (Best Quality)

Use this for most documents, especially invoices with complex layouts.

python run_pipeline.py

• Input: Images in input/ folder.
• Output: output/Extracted_Data_Vision.xlsx

Option 2: Legacy OCR (Offline)

Use this if you don't have an API key or need offline capability.

python run_pipeline.py --method ocr

• Input: Images in input/ folder.
• Output: output/Extracted_Data_OCR.xlsx

Project Structure

├── input/                  # Place source images here
├── output/                 # Generated Excel files appear here
├── ocr_data/               # Intermediate OCR JSON data (Legacy Mode)
├── vision_data/            # Intermediate Vision JSON data (Vision Mode)
├── run_pipeline.py         # Main entry point script
├── vision_processor.py     # Llama Vision integration
├── ocr_extraction.py       # EasyOCR integration
├── json_to_excel.py        # Vision JSON -> Excel converter
├── convert_to_excel.py     # OCR JSON -> Excel converter
└── spatial_table_extractor.py # Table logic for Legacy OCR