Open chashimo opened 4 months ago
What do you mean? This model does only image-to-text, there are no bounding boxes to speak about.
I think what he means is the bounding box of the text, something like text detection rather than image-to-text
@chashimo this is not the purpose of this repo. You can use Comic Text Detector for getting bounding boxes (also, that's 80% lighter than Manga OCR model)
Is it possible to get bounding boxes from OCR results?