✍️ Handwritten Text OCR Pipeline

Extract text from handwritten images with advanced preprocessing, OCR, and text cleaning.
A robust pipeline utilizing OpenCV, Tesseract, and FastAPI for seamless image-to-text conversion.

• OCR Training and Inferencing Demo (Keras)☝️

🚀 Features

• Image Preprocessing: Grayscale conversion, denoising, binarization, and skew correction.
• OCR Extraction: Supports multiple languages via Tesseract OCR.
• Text Cleaning: Removes noise, fixes common OCR errors, and structures text.
• FastAPI Integration: REST API endpoint for easy integration into applications.
• Batch Processing: Process all images in a directory automatically.

📦 Installation

Prerequisites

1. Install Tesseract OCR (v5.3.0+ recommended):

   # Linux
   sudo apt-get install tesseract-ocr libtesseract-dev
   
   # macOS
   brew install tesseract
   
   # Windows: Download from [Tesseract installer](https://github.com/UB-Mannheim/tesseract/wiki)

   Ensure Tesseract is added to your system's PATH.

2. Clone the repository:

   git clone https://github.com/yourusername/handwritten-ocr-pipeline.git
   cd handwritten-ocr-pipeline

3. Install Python dependencies:

   pip install -r requirements.txt

Project Structure

your-project-name/
├── images_raw/          # Raw input images
├── images_processed/    # Preprocessed images
├── ocr_output_raw/      # Raw OCR results
├── ocr_output_processed/ # OCR results from processed images
├── cleaned_text/        # Final cleaned text
├── temp_uploads/        # Temporary API uploads
└── src/                 # Source code

🛠 Usage

1. Run the Main Pipeline

1. Place images in images_raw/ (e.g., image_raw_1.png).

2. Execute the pipeline:

   python src/main_pipeline.py

3. Check outputs in: images_processed/: Preprocessed images. cleaned_text/: Final cleaned text.

2. Use the FastAPI Endpoint

1. Start the API server:

   uvicorn src.api:app --reload

2. Visit http://localhost:8000/docs to access interactive Swagger documentation
3. Upload an image via the /extract-text/ endpoint to receive JSON-formatted text.

🔧 Customization

• Preprocessing: Adjust parameters in src/preprocess.py (e.g., blur kernel size, binarization method).
• OCR Language: Modify the lang parameter in ocr_extract.py (e.g., lang='fra' for French).
• Text Cleaning: Add custom regex rules in text_clean.py to fix domain-specific errors.

🐛 Troubleshooting

• Tesseract Not Found: Set the Tesseract path explicitly in ocr_extract.py:

pytesseract.pytesseract.tesseract_cmd = r'/usr/bin/tesseract'  # Linux/macOS example

• Image Skew Issues: Tweak the Radon transform parameters in correct_skew() (requires scikit-image).

🤝 Contributing

Contributions are welcome! Open an issue or submit a PR for:

• Improved skew correction algorithms
• Additional language support
• Enhanced text structuring rules