Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

In my experience Tesseract improves massively if you can identify the font the text is written in and prepare a custom trained dataset for it to use. See https://github.com/tesseract-ocr/tesseract/wiki/TrainingTess...



All of the failures were directly related to improperly isolated input. In addition, a huge percentage of Chinese text is written in very few fonts.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: