Deepseek-ocr cant read top right corner

#101
by Bunkid - opened

DeepSeek OCR fails to extract text from the top-right section of the input page. Other areas are processed correctly, but the top-right region is either skipped or returns empty results.
I need to get the invoice number and series.
What I’ve tried:

Testing with multiple PDFs/images

Resizing and enhancing image quality

None of these approaches fixed the issue. Would appreciate any help, workarounds, or insights.
image

Are you rasterizing images from pdf and applying any kind of resizing?

Sign up or log in to comment