Recognize text in scanned PDFs
Turn a scan or a photo into selectable text without sending the document to a cloud OCR service.
What it does
The tool renders the PDF pages or opens the image and runs Tesseract.js in the browser. You can choose the language, the reading quality and the pages to process.
The result can be copied as text, downloaded as TXT or turned into a PDF with a selectable text layer over the original image. On first use, the language model (a few MB) is downloaded and cached by the browser.
How to use it
- Open OCR PDF.
- Add a scanned PDF or an image.
- Choose the document language and the reading quality.
- Limit the pages if the document is long.
- Press “Recognize text” and copy the text or download the TXT or the PDF with a text layer.
Why ConvUp
Selectable text
Copy, search and edit the content of a scan.
Cached model
Only the first run downloads the language; it then works offline.
No document upload
Tesseract.js runs locally; the file is not sent.
When OCR fails and how to improve it
OCR works when the page is straight, well lit and printed. It fails with skewed photos, shadows, handwriting, decorative typefaces or scans below 200 DPI. If the output comes back with broken words, rescan at 300 DPI in greyscale and crop the black borders before trying again.
Tables are another weak spot: OCR returns text in the order it finds it, not in rows and columns. For tabular data use PDF Tables to CSV, which tries to rebuild the grid, and always review the output before importing it into a spreadsheet.
Frequently asked questions
Is my document sent to a server?
No. OCR runs in the browser. The only external request is the language model download from tessdata.projectnaptha.com, which does not include your document.
Which languages can I use?
The languages available in Tesseract, including English, Spanish and other common ones. Pick the right language to improve accuracy.
Why are there mistakes in the result?
Accuracy depends on scan quality, resolution, contrast and skew. Try the highest quality and crop or straighten the image before uploading it.
Can I search the text in the resulting PDF?
Yes. The PDF-with-text-layer option places the recognized text over the image so you can select and search it.