Skip to main content

Extract Text

Extract Text reads the text in a photo. It uses Tesseract on the backend CPU.

Use it​

  1. Open a photo.
  2. Expand Extract Text. It is below Common Metadata.
  3. Click Recognize Text.
  4. Edit the text if needed.
  5. Click Save.

Opening the panel shows saved text only. It does not start recognition. Photoly saves a result only after a successful run.

Copy copies the current draft, even before you save. If the clipboard fails, the text stays selected so you can copy it by hand.

Recognize again​

Recognize Again asks before it replaces saved text and drafts. If recognition fails, Photoly keeps both.

Unsaved changes​

Photoly asks you to save, discard or cancel when you have unsaved text and you:

  • Close the panel.
  • Open a similar image.
  • Return through viewer history.
  • Enter the image editor.

If a save fails, the draft stays. On browser unload, the browser shows its own prompt. When you log out, Photoly clears local state and cancels client requests.

Privacy and scope​

  • Only the photo owner can see the text. This is true even when the image is shared.
  • OCR does not need AI search. It does not load embedding models.
  • Not included: video OCR, batch processing, bounding boxes and search integration.

For deployment, limits and API, see Extract Text operations.