Guides › How to recognize a table from an image
How to recognize a table from an image
A practical guide to local OCR in the browser.
Step by step
Start with a clear, high-resolution image. Choose the language matching the document, correct rotation and contrast, then run OCR. Review low-confidence words before exporting the result.
OCR quality depends on print, resolution, lighting and page layout. Local processing improves privacy, but it cannot recover details that are missing from the source.
For scanned PDFs, pages are first rendered locally as images. A PDF with an existing text layer should normally use direct extraction instead of OCR.
Common problems
- Wrong OCR language.
- Text is too small or blurred.
- Page is rotated or has uneven lighting.
- A complex multi-column layout needs manual regions.