docling-project / docling-project/docling
VLM as OCR engine
Open
question
- Dominant language
- Python
- Stars
- 66.4k
- Forks
- 4.8k
- Avg merge
- 2d 21h
- Merged PRs (30d)
- 84
Description
I would like to use a VLM as an OCR engine. What I envision that normal text in pdfs is converted using the deterministic backend, while images are described using a VLM. Is this currently possible?
If I understand the documentation correctly, if the VLM pipeline is configured the entire PDF is always sent to the VLM, even if the pdf consists of structured text. Is that correct?
Contributor guide
Assessment
This issue has not been assessed yet.