Handwritten Text Recognition with EasyOCR and TrOCR

This project is a handwritten text recognition application that allows users to draw on a canvas and detect the handwritten text using OCR models. The project uses EasyOCR and TrOCR for text recognition.

Project Structure

Files

• canvas_widget.py: Contains the CanvasApp class for drawing on the canvas.
• gui.py: Contains the DrawingApp class for the main GUI application.
• main.py: Entry point of the application.
• ocr_processor.py: Contains the OCRProcessor class for processing OCR text.
• trocr_model.py: Contains the TrOCRModel class for TrOCR model integration.

Installation

1. Clone the repository:

   git clone <https://github.com/m01ali/OCR-Handwriting.git>
   cd <repository-directory>

2. Create a virtual environment:

   python -m venv venv

3. Activate the virtual environment:

   • On Windows:

     venv\Scripts\activate

   • On macOS/Linux:

     source venv/bin/activate

4. Install the required packages:

   pip install -r requirements.txt

Usage

1. Run the application:

   python main.py

2. Use the canvas to draw handwritten text.

3. Click the "Detect Text" button to recognize the handwritten text.

4. Two images will be displayed in the workspace files:

• output.png: captures the output (word you wrote on canvas)
• enhanced_image.png: enhances the output image for the OCR model (This step has been added for better debugging purposes)

OCR Models

• EasyOCR: Used for general OCR tasks.
• TrOCR: Used for handwritten text recognition.

Acknowledgements

• EasyOCR
• TrOCR

License

This project is licensed under the MIT License