ortho-editor 0.1.5

ORTHO mask editor tool

ORTHO Mask Editor

A PyQT5-based mask editor tool for the ORTHO (Old Research Text Hebrew OCR) project. This tool helps create training data for the ORTHO U-Net AI model by allowing users to create and edit image masks of Hebrew text documents.

About ORTHO

ORTHO (Old Research Text Hebrew OCR) is a project focused on optical character recognition for historical Hebrew texts. The project uses a U-Net architecture for text segmentation and recognition.

Purpose

This mask editor serves as a crucial tool in the ORTHO pipeline by:

• Creating binary masks for training data
• Marking text regions in historical Hebrew documents
• Preparing ground truth data for the U-Net model
• Enabling manual correction of automated segmentation results

Installation

Prerequisites

• Python 3.6 or higher
• PyQt5
• NumPy

Setup

1. Clone the repository:

git clone https://github.com/yaacov/ortho-editor.git
cd ortho-editor

Usage

1. Prepare a directory that contains two subdirectories named inputs and targets.

   • inputs holds the original images.
   • targets holds the corresponding mask images (same filenames, same dimensions).

2. Install the required dependencies:

   pip install -r requirements.txt
   

3. Run the application:

   python main.py
   

4. When prompted, select your directory with inputs and targets subdirectories.

5. A list of files from the inputs folder will appear. Choose a file to display the image as the background, then draw or edit the mask on top.

License

MIT License