What does this PDF OCR tool do?
Open one PDF in the browser to check each page for selectable text. The page list labels pages as Text layer detected or Image-based, and shows the existing character count. Image-based pages are selected for OCR by default. You can export the recognized text or create a searchable PDF.
Choose which pages to OCR
Use Which pages to OCR to set the selection mode:
- Only image-based pages (recommended) selects pages without an existing text layer.
- Force OCR on every page in range selects every page in the range, including pages with a text layer. Checked pages are OCR'd and re-rendered in the searchable-PDF export.
- Use my manual selection below only keeps the current checkbox selection unchanged.
Use the page checkboxes to adjust the queue. You can also use Select image-based pages, Select all in range, or Clear selection.
Set language, rotation and page range
- Language(s) is a multi-select control with English selected by default. Use Ctrl/Cmd-click to choose more than one language. The selected languages apply to every page selected for OCR, including text-layer pages when Force OCR on every page in range is enabled.
- Rotation for OCR offers None (0°), Rotate 90° clockwise, Rotate 180°, and Rotate 270° clockwise (90° counter-clockwise). It affects OCR reading and the rendered OCR page image.
- Page range accepts comma-separated pages and ranges such as
1-3,7,10-12. Leave it blank to process every page. Invalid or out-of-range entries are ignored. OCR and extracted-text processing apply only to pages in the range; pages outside it are copied in the searchable-PDF output.
Review results and OCR errors
OCR starts automatically after a usable PDF is loaded and whenever its options or page selection changes. The results table reports the page, source, confidence, character count, and status.
- For OCR pages, confidence is the page-level value returned by Tesseract and displayed as a percentage.
- Text-layer results that were not OCR'd show 100% confidence.
- Skipped pages have no confidence value.
- OCR errors appear with the affected page number and error message.
Export extracted text or a searchable PDF
- Download extracted text (.txt) exports only pages in the selected range, with markers identifying OCR, text layer, skipped, or error results. Pages outside the range are not included.
- Download searchable PDF creates a new PDF with the same page count and order. OCR-selected pages are re-rendered, encoded as JPEG images, and given an invisible OCR text layer. They do not preserve the exact original page image or an existing text layer unchanged.
Pages not selected for OCR, including pages outside the range, are copied from the source PDF. If OCR fails for a selected page, the searchable-PDF export copies that page without an OCR text layer. Rotation can change the rendered page orientation and dimensions. Unsupported characters may be omitted from the invisible text layer, while the rendered page image remains.
Limitations and local processing
- The drop zone accepts one PDF file at a time.
- PDFs with more than 300 pages are rejected.
- Password-protected or encrypted PDFs are refused during loading.
- The tool recognizes plain running text, not table rows, columns, headers, or merged-cell structure. Use the separate "PDF & Image Table Extractor" for table structure.
- Processing and export occur in the browser, and this tool does not upload the PDF.
- Unreadable or malformed PDFs can show Could not open this PDF. If a required browser library is unavailable, processing can stop with PDF library failed to load., OCR library failed to load., or PDF export library failed to load.
How to OCR a PDF
- Drop a PDF file onto the drop zone, or click browse file.
- To replace the current PDF, click Change file, then use the drop zone or browse file to load the replacement.
- Set language, rotation and page range under OCR options.
- Review the per-page list, including existing character counts. Image-based pages are pre-selected by default; adjust the selection or change the mode if needed.
- Wait for automatic processing to finish, then review the per-page progress and results.
- Check the results table for confidence and any errors, then download the extracted text or the searchable PDF.
Frequently asked questions
Does this tool OCR every page?
Not by default. The default mode selects image-based pages for OCR. You can select pages manually, or use Force OCR on every page in range to OCR checked text-layer pages too. OCR-selected pages are re-rendered in the searchable-PDF export.
Will the searchable PDF output look different from my original?
It can. OCR-selected pages are rendered again, encoded as JPEG images, and placed in a new PDF with an invisible OCR text layer. Their rendering may differ from the source, especially when rotation is applied. Pages not selected for OCR, including pages outside the range, are copied from the source PDF.
Does this tool reconstruct tables?
No. This tool recognizes plain running text only. For reconstructing table rows, columns and merged cells from a PDF or image, use the separate "PDF & Image Table Extractor" - presenting table layout as generic OCR text would lose the row/column structure.
What happens with password-protected PDFs?
The tool does not attempt to bypass encryption. If a PDF is password-protected, loading fails with a clear error and no processing happens. Remove the password with the file's owner permission first, in a tool designed for that, then re-open it here.
Does this tool upload my file anywhere?
The PDF is read and processed in your browser with pdf.js, Tesseract.js and pdf-lib. This tool does not upload the PDF.