search

Danily07 / Translumo

Advanced real-time screen translator for games, hardcoded subtitles in videos, static text and etc.
Apache License 2.0
3.3k stars 162 forks source link

Image preprocessing #4

Open zamazan4ik opened 1 year ago

zamazan4ik commented 1 year ago

Hi. Thanks for the project!

A few years ago I developed an in-house ad-hoc OCR solution based on Tesseract. And want to share some suggestions. As you probably already know, Tesseract has very weak preprocessing phase (not-so-advanced binarization techniques, trash removal, etc.). To help with it, I developed a library with pre-recognition algorithms: https://github.com/zamazan4ik/PRLib Maybe you would be able to find something useful there.

Danily07 commented 1 year ago

Hi! Thanks for sharing the solution! The thing is the application needs as universal pre-processing as possible for different types of images (cos I don't want delegate manual adjustment to end-users). So, yeah, now the implementation has quite weak pre-processing for Tesseract and I defenitely have plans to consider improvements in that way.