This is a self contained package of the Deep Optical Measure Detector. It can be utilized to generate measure annotations in the MEI format (compatible to v3 & v4) for handwritten and typeset score images.
This code was meant as a quick functional demonstration. It is not production ready or even documented! Handle with care.
-
Make sure to have Docker installed and running properly with at least 4 GB of RAM assigned to Docker.
-
Run the container in a terminal:
docker run -p 8000:8000 -i --rm cemfi/measure-detector- Go to http://localhost:8000 and drop some images. Be patient, the detection is computationally pretty heavy.
The DNN model was trained by Alexander Pacha, see this project. Thanks also to Alexander Leemhuis for meticulously annotating hundreds of score images for the dataset.