Line layout analysis algorithms for the OMMR4all project. The tool uses the output of https://gitlab2.informatik.uni-wuerzburg.de/OMMR4all/ommr4all-line-detection to segment the provided image in several regions.

Dependencies for the project are located in requirement.txt.
Major Dependencies are:

• numpy
• scipy
• pillow
• opencv

This projects requires the line detection of https://gitlab2.informatik.uni-wuerzburg.de/OMMR4all/ommr4all-line-detection to be installed locally.

• Clone the line segmentation repository git clone https://gitlab2.informatik.uni-wuerzburg.de/OMMR4all/ommr4all-line-detection

• (Optional but recommended) Activate your virtual environment (adapt to your path): source venv/bin/activate

• install line segmentation cd page-ommr4all-line-detection && python setup.py install

• Since the line detection requires the Pixel classifier of CWick: https://gitlab2.informatik.uni-wuerzburg.de/chw71yx/page-segmentation it is recommend to install it

• Clone the page segmentation repository git clone https://gitlab2.informatik.uni-wuerzburg.de/chw71yx/page-segmentation

• install page segmentation cd page-segmentation && python setup.py install

• clone the layout analysis git clone https://gitlab2.informatik.uni-wuerzburg.de/OMMR4all/ommr4all-layout-analysis

• install the specific requirements of layout analysis cd ommr4all-layout-analysis && pip install -r requirements.txt or cd ommr4all-layout-analysis && python setup.py install to install the provided scripts, too

• the line analysis is installed

If you simply want to use the tool to your images you can use the scripts provided by the tool.

Note that you have to activate the virtual environment if used during the installation in order to make the command line scripts available. You can segment the regions in the image by calling:

layout-analysis-predict --gray your_gray_scale_images.*.png

Note that there are several important and optional parameters to adjust and enhance the prediction.

For a full list type layout-analysis-predict --help

Input	Predicted Regions

• Alexander Hartelt
• Christoph Wick