Skip to content

Comment on Frog: OCR Tool for Linuxparent

Comments

Working for a library on rare fonts I've found Tesseract fantastic for custom training.

It certainly beats Abbyy from 10 years ago - maybe a low bar to clear.

I had to spend some time setting up labeling then did some supplemental training on UB-Mannheim datasets.

Tesseract is the only OCR FOSS solution that has reasonable performance.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.