OCR

Reads the text off a scan or a screenshot so it can be searched.

The OCR node extracts text from images and document scans using Tesseract (via Tess4J). The recognized text is stored so it can be indexed and searched.

Kind

ocr

Applies to

Image

Input ports

media

Output ports

text (String)

Requirements

Tesseract / Tess4J plus its tessdata language files on the worker. CPU-bound; storage for the tessdata models.

Persists to

asset_json_comp + asset_node_result ledger

Configuration

Figure 1. The node’s settings in the pipeline editor

Set these in the panel above, or in the node’s options block in a pipeline definition:

Option Meaning

Option	Meaning
`tessDataPath`	Path to the Tesseract `tessdata` directory
`language`	Recognition language(s), e.g. `eng`, `deu`

tessDataPath

Path to the Tesseract tessdata directory

language

Recognition language(s), e.g. eng, deu

Seeing it run

Turn on Debug Mode and every node keeps what it produced, on the card itself. Below is a real run of this node over albert_einstein.png.

The strip on the card lists what each output port carried — text.

Use Cases

Searchable scans — index text from screenshots, receipts and scanned documents.
Document workflows — surface embedded text for classification or routing.
Compliance — find images that contain particular text.

Looking for something else?

All documentation Feature overview