@markitdownjs/image-ocr

Image OCR converter for MarkItDownJS. Extracts text from images using Tesseract.js.

Install

npm install @markitdownjs/image-ocr @markitdownjs/core

# Peer dependency (optional — required at runtime)
npm install tesseract.js

Peer dependency: tesseract.js >= 5.0.0 must be installed separately. The converter will throw at runtime if it is missing.

Usage

import { MarkItDown } from "@markitdownjs/core";
import { OcrConverter } from "@markitdownjs/image-ocr";

const parser = new MarkItDown();
parser.registerConverter(new OcrConverter({ lang: "eng" }));

const result = await parser.convert({ source: imageBuffer, mimeType: "image/png" });

API

OcrConverter

Implements the IConverter interface from @markitdownjs/core.

Constructor options:

| Option | Type | Default | Description | |--------|------|---------|-------------| | lang | string | "eng" | Tesseract language code | | workerPath | string | — | Custom path to Tesseract worker |

| Method | Description | |--------|-------------| | convert(input) | Runs OCR on the image buffer and returns extracted text as document nodes | | canHandle(mimeType) | Returns true for image/png, image/jpeg, image/webp, image/tiff |

Part of the MarkItDownJS Monorepo

https://github.com/markitdownjs/markitdownjs