Professional OCR Repository

This repository contains a clean, production-ready set of OCR (Optical Character Recognition) tools designed for high-quality text extraction from PDFs and images, with a focus on resumes and structured documents.

🚀 Features

• Multiple OCR Engines: Choose between different state-of-the-art OCR technologies.
• Hybrid Extraction: Combines advanced image preprocessing (OpenCV) with Docling for superior accuracy on noisy documents.
• Automated Cleaning: Built-in post-processing to remove OCR artifacts, fix common misreads, and normalize spacing.
• Text Optimization: Intelligent logic to detect and reposition text blocks for accurate extraction.
• GPU Support: Automatic GPU acceleration for faster processing when available.

---

🛠️ Available OCR Engines

1. Docling OCR (docling)

• Base: IBM Docling
• Best for: Standard digital and high-quality scanned PDFs.
• Output: Clean Markdown with preserved table structures.

2. PaddleOCR (paddle)

• Base: PaddleOCR
• Best for: Multilingual support and documents with complex layouts or handwriting.
• Process: Includes custom image preprocessing, de-skewing, and line merging.

3. Hybrid Engine (hybrid)

• Process: PyMuPDF (fitz) -> OpenCV Preprocessing -> Docling OCR
• Best for: Low-quality scans, noisy backgrounds, or documents that fail with standard methods.
• Technique: Uses CLAHE (Contrast Limited Adaptive Histogram Equalization), adaptive thresholding, and denoising before extraction.

---

📦 Installation

1. Clone the repository:

   git clone <repo-url>
   cd Multi-Engine-OCR

2. Install dependencies:

   pip install -r requirements.txt

   Note: Ensure you have Tesseract and/or PaddlePaddle installed depending on your OS requirements.

---

🏃 Usage

You can run any OCR engine using the run.py script:

# Basic usage (defaults to Docling)
python run.py path/to/your/document.pdf

# Using PaddleOCR
python run.py path/to/your/document.pdf --engine paddle

# Using the Hybrid Engine
python run.py path/to/your/document.pdf --engine hybrid

# Specify output directory
python run.py document.pdf --output my_results

# Force CPU usage
python run.py document.pdf --no-gpu

---

📁 Repository Structure

.
├── engines/
│   ├── docling_engine.py   # Docling implementation
│   ├── paddle_engine.py    # PaddleOCR implementation
│   └── hybrid_engine.py    # Image + Docling hybrid implementation
├── utils/
│   └── text_processing.py  # Cleaning & post-processing logic
├── run.py                  # Main entry point CLI
├── README.md               # Documentation
└── requirements.txt        # Project dependencies

---

Meet the Developer

📧 Email: hassanaiengineer@gmail.com
🔗 LinkedIn: Hassan Khan
🔗 Upwork: Hassan Khan on Upwork

---

Made with ❤️ by Hassan Khan