ocrd_kraken is an incomplete implementation. It only contains binarization which is based on ocropy and which is implemented better and more flexibly in ocrd_cis. The interesting part, the text recognition and baseline layout detection, is missing.
How it should be
Users should be able to use ocrd-kraken-recognize to run text recognition and ocrd-kraken-segment to run layout detection with kraken.
Steps
[ ] finish PR to implement kraken recognition
[ ] remove binarization code
[ ] Update workflow guide to include new processors, remove binarization
Current situation
ocrd_kraken is an incomplete implementation. It only contains binarization which is based on ocropy and which is implemented better and more flexibly in
ocrd_cis. The interesting part, the text recognition and baseline layout detection, is missing.How it should be
Users should be able to use
ocrd-kraken-recognizeto run text recognition andocrd-kraken-segmentto run layout detection with kraken.Steps