You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Go to file
vahidrezanezhad f3c6ac8d70
Update test.yml
4 weeks ago
.circleci make git submodule update depend on subdir 2 months ago
.github/workflows Update test.yml 4 weeks ago
repo add assets subrepo 3 years ago
sbb_binarize 📦 v0.1.0 2 months ago
.gitignore 📦 v0.0.2 3 years ago
.gitkeep Add new directory, you can find corresponding models in qurator-data 4 years ago
.gitmodules add assets subrepo 3 years ago 📦 v0.1.0 2 months ago
LICENSE Add LICENSE 4 years ago
Makefile make git submodule update depend on subdir 2 months ago Update 2 months ago Add new file 4 years ago
ocrd-tool.json add ocrd-tool.json 3 years ago
requirements.txt Update requirements.txt 8 months ago fix standalone CLI version_option 2 months ago


Document Image Binarization using pre-trained models

pip release CircleCI test GHActions Tests



Python versions 3.7-3.10 are currently supported.

You can either install via

pip install sbb-binarization

or clone the repository, enter it and install (editable) with

git clone
cd sbb_binarization; pip install -e .


Pre-trained models can be downloaded from the locations below. We also provide the models and model card on 🤗

Version Format Download
2021-03-09 SavedModel
2021-03-09 HDF5
2020-01-16 SavedModel
2020-01-16 HDF5

With OCR-D, you can use the Resource Manager to deploy models, e.g.

ocrd resmgr download ocrd-sbb-binarize "*"


sbb_binarize \
  -m <path to directory containing model files> \
  <input image> \
  <output image>

Note: the output image MUST use either .tif or .png as file extension to produce a binary image. Input images can also be JPEG.

Images containing a lot of border noise (black pixels) should be cropped beforehand to improve the quality of results.


sbb_binarize -m /path/to/model/ myimage.tif myimage-bin.tif

To use the OCR-D interface:

ocrd-sbb-binarize -I INPUT_FILE_GRP -O OCR-D-IMG-BIN -P model default


For simple smoke tests, the following will

  • download models

  • download test data

  • run the OCR-D wrapper (on page and region level):

      make models
      make test