Language Support

[balaatsky]

Hi, Thank you for this good work. Can we get the language support and other ideas from: https://github.com/JaidedAI/EasyOCR/tree/master/easyocr/character ?

[robertknight]

This issue needs a more specific scope, otherwise it is hard to say when it is "done". Do you have a particular use case in mind?

The link you shared is a set of text files that contain alphabets for different languages. In this project you can already specify the alphabet via OcrEngineParams::alphabet when using ocrs as a library or --alphabet in the CLI. In order to actually recognize text in other languages, training datasets need to be created and models need to be trained. See https://github.com/robertknight/ocrs-models. Then the appropriate alphabet needs to be specified at inference time.