A computer vision and text matching-based tool for automatically identifying and annotating errors and highlights in handwritten text.

The image above shows an annotated handwritten essay with error markers (red underlines) and highlights (green backgrounds).

• OCR-based handwritten text recognition
• Fuzzy text matching algorithm for improved recognition accuracy
• Intelligent annotation system with error marking and content highlighting
• Dynamic text position adjustment to avoid annotation overlap
• API interface support for easy integration into other systems

• Python
• OpenCV
• FastAPI
• PIL (Python Imaging Library)
• FuzzyWuzzy (Text Matching)

1. Clone the repository

git clone [repository-url]
cd handwritten-essay-annotator

2. Install dependencies

pip install -r requirements.txt

3. Set up configuration

cp config.py.example config.py
# Edit config.py and fill in your Baidu OCR API credentials
# Get your API credentials at: https://cloud.baidu.com/product/ocr

The project provides two versions of the annotation service:

Option 1: Modular Version (Recommended)

# Runs on port 8006
python main1.py

Option 2: Standalone Version

# Runs on port 8005
python main.py

# Test the annotation service (connects to port 8006 by default)
python api_test.py

# Runs on port 8002 - for image rotation and alteration removal
python image_process.py

You need to provide:

• Handwritten text image for annotation
• Reference text for text matching
  • Error text snippets
  • Error types
  • Correction explanations
  • Highlight content markers

• Annotated text image with:
  • Error markers (with numbering)
  • Correction explanations
  • Content highlights

Before using this tool, you need to configure your Baidu OCR API credentials:

1. Copy config.py.example to config.py
2. Register for a Baidu Cloud account and create an OCR application at https://cloud.baidu.com/product/ocr
3. Fill in your client_id and client_secret in config.py

The default ports used by different services:

• main.py: Port 8005
• main1.py: Port 8006 (default for api_test.py)
• image_process.py: Port 8002

You can modify the ports in each file's if __name__ == '__main__': section if needed.

The examples/ folder contains sample images.

MIT

Issues and Pull Requests are welcome to help improve this project.