Add possibility to deactivate OCR #2467
Labels
enhancement
New feature or request
Comments
|
Yes, not only is OCR super slow, but table contents are frequently mangled because This seems like an important problem for people who care about the quality of their data (so, I guess, not anyone using Generative AI, but...) The Aryn partitioner does not insist on OCRing everything, but unfortunately it is tightly bound to a different framework which is not as nice as Unstructured for casual use cases... |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Is your feature request related to a problem? Please describe.
I am working on large text-based PDF files, and would like to parse them as fats as possible, while keeping a high resolution (
strategy="hi_res"). I am interested in extracting tables, and tables from pictures, however I would like to deactivate OCR for images detected as such. The first obvious reason is speed. But also images do not matter much.Describe the solution you'd like
A keyword argument that would enable or disable OCR would be the most easy thing to code I guess and would be a nice additional feature, especially if it can differentiate between tables and images. Another nice feature would be to not perform OCR on tables for text-based regions, in order to speed the partition process.
Describe alternatives you've considered
Adding a new
OCRMode.NO_OCRAdditional context
I am not interested in images s.a. graphs or photos.
The text was updated successfully, but these errors were encountered: