MoPDF icon

MoPDF

Extract text from PDFs, convert pages to images and generate AI-ready output — without any API, running entirely local inside n8n.

Actions4

OCR

AI-generated

Summary

Perform OCR text recognition on an input image binary data using Tesseract.js with configurable languages, page segmentation modes, and output formats.

Inputs

  • Languages — One or more languages for the OCR recognition, e.g. German and English. Multiple languages may slow down processing.
  • Page Segmentation — How Tesseract analyzes the image, such as Automatic (recommended), Single Text Block, Single Line, Single Word, or Sparse Text.
  • Text Output Format — Format of the recognized text output, either Plain Text or Markdown with hOCR structuring (useful for tables).
  • Output Mode — How the OCR text results are output: Per Image (one item per image) or Combined Text (one item combining all images).
  • Image Separator — String separator between individual image results when using Combined Text output mode.
  • Additional Options — Extra OCR options including returning words or lines with coordinates, hOCR HTML output for table recognition, minimum confidence threshold to filter words, and Text Only output without metadata.
  • Input Binary Property — The binary property name of the item containing the image data to be processed.
  • Keep Input Data — Whether to retain the original JSON data from the input item in the output result.

Output shape

a list of items containing OCR results either per image or combined text depending on output mode

Each output item contains recognized text under 'ocrText'. Additional fields may include OCR confidence, languages, page segmentation mode, processing time, and optionally lines, words, or hOCR data depending on selected options. Item pairing preserves input item reference. When output mode is Combined Text, a single output item aggregates text from all input items, including summary metadata if 'Keep Input Data' is enabled. Errors per item are handled according to node error settings.

Examples

Example 1: Run OCR on PNG image binary input field 'data' with German and English languages, output one item per image with plain text.

Set Languages to German and English, Page Segmentation Mode to Automatic, Text Output Format to Plain Text, Output Mode to Per Image, Input Binary Property to 'data', Keep Input Data enabled.

Example 2: Run OCR on multiple images combining text in Markdown format with hOCR enabled for structured table output.

Set Text Output Format to Markdown, Output Mode to Combined Text, Additional Options include enabling hOCR output, and specify an Image Separator string.

Links

Discussion