Find answers from the community

Updated 6 months ago

Does the PDF reader use OCR

At a glance

The post asks if the PDF reader uses OCR. The comments indicate that the PDF reader currently uses a basic pdf2text parser, not OCR. Community members suggest using the donut model, which seems to be effective for processing images, but it is not clear if it can be used for parsing PDFs. Some community members express interest in having a donut-based PDF parser as a simpler alternative to the current solution.

Useful resources

vvkdi5cord

Does the PDF reader use OCR?

7 comments

jjerryjliu0

not yet, it uses a basic pdf2text parser

SSJ

@conceptofmind in LangChain discord likes the following: https://github.com/clovaai/donut

SSJ

which looks like it should do wonders... https://towardsdatascience.com/ocr-free-document-understanding-with-donut-1acfbdf099be

jjerryjliu0

we use the donut model to parse images!

jjerryjliu0

so i amend my previous statement. the image parser uses the donut model, the pdf parser does not

rrtk

Sorry for the late response, but is there a reason you don't use the donut model to parse PDFs? And is there a simpler alternative for PDFs?

jjerryjliu0

we have a donut model to process images, but if you want to add a donut pdf parser that would be appreciated too!

Add a reply