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

Robert Sachunsky 0b3d07a4c4 add badges		2 years ago
.circleci	CircleCI: try to fix syntax	2 years ago
.github/workflows	GHA: clone submodules	2 years ago
repo	add assets subrepo	4 years ago
sbb_binarize	fix region level	2 years ago
.gitignore	📦 v0.0.2	4 years ago
.gitkeep	Add new directory, you can find corresponding models in qurator-data	5 years ago
.gitmodules	add assets subrepo	4 years ago
CHANGELOG.md	📦 v0.0.11	2 years ago
LICENSE	Add LICENSE	5 years ago
Makefile	Update Makefile	2 years ago
README.md	add badges	2 years ago
make.sh	Add new file	5 years ago
ocrd-tool.json	add ocrd-tool.json	4 years ago
requirements.txt	Update requirements.txt	2 years ago
setup.py	minimal CI setup	4 years ago

Binarization

Binarization for document images

Examples

This tool performs document image binarization using a trained ResNet50-UNet model.

Clone the repository, enter it and run

pip install .

Pre-trained models in HDF5 format can be downloaded from here:

We also provide models in Tensorflow SavedModel format via Huggingface and Github release assets:

With OCR-D, you can use the [Resource Manager](Tensorflow SavedModel) to deploy models, e.g.

ocrd resmgr download ocrd-sbb-binarize "*"

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

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 model
  make test
```