ocropus-models

Here are some of my models I created for text recognition with OCRopus https://github.com/tmbdev/ocropy

sh1890.pyrnn.gz

German book from the end of 19th century ("Ortschafts-Verzeichnis für die Provinz Schleswig-Holstein"). The complete book can be found here: http://dibiki.ub.uni-kiel.de/viewer/resolver?urn=urn:nbn🇩🇪gbv:8:2-2750405 With the model you can achieve an error rate somewhere between 1.5% and 0.7%.

font example for sh1890

verlustliste_österreich-ungarn

I have moved the model and the images used for training to a separate repository: https://github.com/jze/ocropus-model_oesterreich-ungarn