I cant find any OCR which is fast and accurate to an extent where if I have 10000 scanned pdfs (pdfs that have been scanned. They are scanned from mobile)
I have tried various vision language models like PaddleOCR VL pipeline, also used some other things which i got. Though they are nearly accurate.. they are painfully slow.
I have a very solid gpu. RTX 6000 pro blackwell.
So what can i run which can be blazinggly fast and also accurate at same time
[link] [comments]