Short answer
Try to select a sentence in the PDF. If you cannot, check the source's Language under Edit Metadata, then install OCR and apply it to the scan. Search for one known phrase, then compare names, numbers, and quotations with the page image.
This guide covers CiteFlow. Check requirements and download options before you begin.
On this page
When to use this method
Use this method when a PDF page is a picture of text instead of selectable text. OCR adds searchable text so you can find words later. Opening or reading a PDF does not start OCR automatically.
Before you start
Open one page and try to select a sentence. OCR turns letters in an image into searchable text. Keep the original intake PDF outside CiteFlow's managed library: successful OCR switches the source to a searchable copy and can remove the previous managed copy. Check the source's Language in Edit Metadata before choosing Install OCR and Apply, which starts recognition after installation. The component supports eight text languages and includes script detection. CiteFlow does not apply automatic page-orientation correction; make sure the page is upright.
Check retrieval, then check the quotation
After recognition, search for a phrase visible on the scanned page. Open that result and compare the words with the image. A found page confirms that some text is searchable; it does not prove that every name, number or symbol was recognized correctly. This guide checks retrieval within the research project. For a separate searchable-PDF output, follow the PDF Tools OCR guide.
Install and verify optional OCR
-
Check whether OCR is needed
Try to select one sentence or search for a phrase visible on the page. If neither works, the page likely needs OCR. Before starting it, open the source menu, choose Edit Metadata and check its Language.
-
Install OCR
Open Settings > Install OCR, or choose Install OCR and Apply on the scanned PDF. Approve the download. You should see progress followed by a ready state. If it fails, retry or repair the component.
Use Install OCR and Apply only when visible text cannot be selected or found. -
Understand the language choice
OCR uses the source Language you checked, available language models and page-level script detection. There is no separate language picker in the progress area. Supported text languages are Simplified Chinese, English, French, German, Japanese, Russian, Spanish and Turkish. The interface language does not select the document language, and script detection does not rotate the page.
-
Apply OCR
If you installed from Settings, use the source's Apply OCR action. Install OCR and Apply continues automatically after a successful install. Watch progress in the project workspace. An OCR-in-progress badge means the work is still running; a completed badge appears only after the searchable copy is ready.
You can cancel OCR from its progress area. If it is canceled or fails before the searchable copy is activated, the previous source remains available and the attempt is not marked complete. Check the reported state before retrying.
-
Check the recognized text
Search for one phrase you can see on the page. You should find that page. Compare names, dates, symbols, tables, and quotations with the image before using them.
OCR installation details
The optional installation downloads about 212 MB and includes the OCR engine, eight text-language models, and the orientation and script model. It installs inside CiteFlow's data folder, verifies the downloaded files, and does not require administrator rights or changes to the Windows PATH.
OCR evidence risks
- Running OCR on a PDF that already has reliable text
- Leaving the wrong document language selected
- Accepting names or numbers without visual comparison
- Discarding the source page image
Checkpoint: Recognition checks
- Verify the page is image-only.
- Confirm the managed component and the selected document language.
- Search one known phrase after OCR.
- Compare every quoted or cited value with the scan.