Docs / Transformers / Extract Text (OCR)

Extract Text (OCR)

Run Apple's Vision OCR on an image and output the recognized text. Runs 100% on-device — no network, no AI provider required.

Because recognition happens entirely on your Mac with Apple's Vision framework, no image or text ever leaves the machine — it fits the local-first model. Extract Text (OCR) pairs naturally with the Photo Added trigger, but works on any image file path in the incoming payload.

Ports

Inlet
Input (image)
Outlet
Output (any)

Configuration

Parameter Type Default Description
imagePath variable picker {{filePath}} The image to read. Defaults to {{filePath}} from the upstream node.
recognitionLevel dropdown Accurate Choose Accurate or Fast. Accurate is slower but reads more reliably.
languageCorrection toggle true Applies dictionary-based correction. Turn off for codes, SKUs, or serial numbers where correction hurts.
outputKey text "extractedText" The payload key where the recognized text is stored.

Output Payload

These keys are merged onto the incoming payload — upstream keys such as assetIdentifier and filePath are preserved, so a later node can still act on the photo.

extractedText The recognized text (or the key name configured via outputKey).
ocrConfidence A number indicating the recognition confidence.
ocrLineCount The number of text lines recognized in the image.

Example