Ask questions about a scanned PDF

What happens when you upload a scanned PDF

1. Your browser checks each page for a text layer. Pages that already have text are left alone.
2. Pages without text are rendered and read with OCR on your own device, so the heavy work does not wait in a server queue. You see a progress bar while it runs.
3. The recognised text and the position of each line are uploaded with the file. If the browser cannot run OCR, the server reads the scanned pages instead.
4. From then on the document works like any other: ask questions, and click a quote to see it highlighted on the scanned page.

Getting good results from a scan

• Scan at 200-300 DPI if you can. Very low-resolution scans produce more recognition errors.
• Straight, evenly lit pages read far better than angled phone photos.
• Check numbers and names against the highlighted page. OCR mistakes in figures carry into answers.
• Handwriting is not reliably recognised; typed and printed text is.

Mixed documents

Many PDFs mix typed pages with scanned ones, like a contract with a scanned signature page or a report with scanned appendices.Docamend checks page by page, so typed pages keep their exact text and only the scanned pages go through OCR.

Limits to know

• OCR currently reads English text. Scans in other languages are not recognised reliably.
• Recognition quality depends on the scan. Faded, skewed or handwritten pages may produce errors or be skipped.
• Browser OCR uses your device's processor, so a long scanned document can take a while on an older phone or laptop.

Questions