In this forum, we will work with people who are training OCR engines for Polytonic Greek, helping them improve the quality of their engines by correcting specific texts.
2 posts • Page 1 of 1
What's the best way to identify the fonts that are used in a scanned image?
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
I should make a field guide to this. For OCR purposes, what matters is the performance of a given engine/classifier combination. Ocropus/Ciaconna, e.g., appears to be more comprehensive, allowing various types of Porson fonts to be classified together, which is a really good thing for commentaries. I'm not sure how far this extends: can I classify all Greek fonts together and not have the results degrade (or even improve)? Our Tesseract classifiers are, I believe, currently all completely comprehensive. Gamera/Rigaudon is much pickier, so one would need to make a more specific field-guide. I think it makes sense to start with the categories that I'm currently building up in Ciaconna (the Ocropus-based material).