metadata
license: gpl
language:
- he
tags:
- hebrew
- ocr
This project aims to train a ORTHO (Old Research Text Hebrew OCR) lines catcher model to perform line tracing tasks on large text images by splitting them into smaller patches. The approach allows for higher-capacity models to handle arbitrarily large text images without running out of memory.
For more information: https://github.com/yaacov/ortho-lines