Image to Text OCR - Extract Text from Images & Scanned PDFs

What does Image to Text OCR do?

Image to Text OCR extracts the text from images and scanned PDFs in bulk. Give it image URLs (screenshots, scanned documents, receipts, invoices, posters, slides, product photos), scanned PDF files, or the image URLs in another Actor's dataset, and get back the text of each image or page with paragraphs and line breaks, a word count, a confidence score and, if you want, every text line with its position.

It runs the open-source Tesseract OCR engine (Apache-2.0) inside the Actor: no Google, OpenAI or other external API, no API key, and your images are not sent to a third party. It reads 12 languages (English, German, French, Spanish, Italian, Portuguese, Dutch, Swedish, Norwegian, Danish, Finnish, Polish), alone or mixed.

Why use Image to Text OCR?

• Bulk OCR without an API key. Hundreds of images in one run, priced per image. No Google Cloud Vision setup, no LLM token bills.
• Scanned PDFs too. Each PDF page is rendered at 200 DPI (configurable) and OCR'd, with the page number in the output.
• Clean text. Paragraphs and line breaks follow the layout of the image. Optional per-line confidence and bounding boxes in original image pixels.
• Tuned for real images. EXIF rotation, transparent PNGs, small screenshots (upscaled before OCR) and huge scans (downscaled) are handled for you. A "sparse" layout mode finds scattered text on photos and posters.
• Chains with other Actors. Pass a dataset ID and the field that holds the image URL, for example the fileUrl from the Bulk Image & File Downloader or imageUrl from a product scraper.
• Private by design. OCR happens inside the Actor run; only the image URLs you give are downloaded, and robots.txt is respected.

What data can you extract?

Field	Example	Description
url	https://tesseract-ocr.github.io/tessdoc/images/eurotext.png	Image or PDF URL
fileType	png / pdf	Image format, or pdf
page / pageCount	2 / 6	Page number and page count (PDFs; pageCount also for multi-frame TIFF/GIF)
text	The (quick) [brown] {fox} jumps!\nOver the $43,456.78 <lazy> #90 dog…	Recognised text, lines joined by \n, paragraphs by a blank line
wordCount / lineCount	66 / 12	Words and lines found
meanConfidence	91.4	Average word confidence (0-100)
languages	["eng"]	OCR languages used
width / height	640 / 500	Image size in pixels
lines	[{"text": "The (quick) [brown] {fox} jumps!", "confidence": 96.1, "box": [65, 41, 450, 30]}]	Optional: every line with confidence and box [x, y, width, height]
processingSeconds	0.59	OCR time for this image or page
foundIn	url / dataset:fileUrl	Where the URL came from
error	http_404 / not_an_image_or_pdf / blocked_by_robots_txt	Why an item was skipped (free)
scrapedAt	2026-10-28T21:12:40Z	Processing time

How to use Image to Text OCR (step by step)

1. Click Try for free.
2. Paste Image URLs and/or Scanned PDF URLs, or pick a Dataset from another run and set its URL field.
3. Choose the Languages in your images (more languages are slower) and the Page layout: Auto for documents and screenshots, Sparse for photos and posters.
4. Optional: set Min word confidence to drop noise, switch on Include lines with boxes, and set Max images / PDF pages.
5. Click Start. Open the Text view or export to JSON, CSV or Excel.

Input example

{
  "imageUrls": [
    "https://tesseract-ocr.github.io/tessdoc/images/eurotext.png"
  ],
  "pdfUrls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf"],
  "languages": ["eng", "deu"],
  "layout": "auto",
  "maxPdfPages": 6,
  "includeLines": false,
  "maxItems": 100
}

Output example

{
  "url": "https://tesseract-ocr.github.io/tessdoc/images/eurotext.png",
  "foundIn": "url",
  "fileType": "png",
  "page": null,
  "pageCount": null,
  "languages": ["eng"],
  "text": "The (quick) [brown] {fox} jumps!\nOver the $43,456.78 <lazy> #90 dog\n& duck/goose, as 12.5% of E-mail\nfrom aspammer@website.com is spam.\nDer ,,schnelle” braune Fuchs springt\n…",
  "wordCount": 66,
  "lineCount": 12,
  "meanConfidence": 91.4,
  "width": 640,
  "height": 500,
  "processingSeconds": 0.59,
  "scrapedAt": "2026-10-28T21:12:40Z"
}

How much does it cost?

Image to Text OCR uses pay-per-event pricing.

Event	Price (Free plan)	When it is charged
Actor start	$0.0005	Once per run
Image OCR	$0.004 ($4 per 1,000 images)	One image processed
PDF page OCR	$0.008 ($8 per 1,000 pages)	One PDF page processed

Images that can't be downloaded, files that are not images or PDFs, files over the size limit and URLs blocked by robots.txt are free. An image is charged once OCR ran, even if it contains no text. Paid Apify plans get Store discounts (Bronze −15%, Silver −30%, Gold and up −45%).

Worked examples (Free plan prices):

• 500 app or website screenshots: ≈ $2.00.
• A 40-page scanned contract: 40 × $0.008 ≈ $0.32.
• 5,000 product photos from a scraper's dataset, to read labels and packaging text: ≈ $20.

A full page of dense text takes 10-25 seconds on the default 1 GB memory; a screenshot with a few lines takes well under a second (the sample image: 0.6 s). More memory gives the run more CPU, so it finishes faster at the same price.

Use cases

• Document digitisation. Turn scanned letters, contracts, forms and archive PDFs into searchable text.
• Receipts and invoices. Pull the raw text of receipts for expense tools or bookkeeping (combine with your own parsing or an LLM for fields).
• Screenshots and social images. Read the text in screenshots, memes, slides and infographics for search, moderation or analytics.
• E-commerce. Read text on product photos and packaging: ingredients, model numbers, labels.
• RAG and AI pipelines. Feed text from image-only documents into vector databases and LLM agents.
• Accessibility. Draft alt texts and transcripts for images that contain text.

Use with AI agents, MCP and the API

curl -X POST "https://api.apify.com/v2/acts/teddythebuilder~image-to-text-ocr/run-sync-get-dataset-items?token=<TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"imageUrls":["https://tesseract-ocr.github.io/tessdoc/images/eurotext.png"],"languages":["eng"]}'

Python with apify-client:

from apify_client import ApifyClient

client = ApifyClient("<TOKEN>")
run = client.actor("teddythebuilder/image-to-text-ocr").call(run_input={
    "pdfUrls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf"],
    "maxPdfPages": 3,
    "languages": ["eng"],
})
for page in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(page.get("page"), page.get("meanConfidence"), (page.get("text") or "")[:200])

MCP: available as a tool in the Apify MCP server (https://mcp.apify.com) for Claude, ChatGPT, Cursor and other MCP clients.

Integrations

Send the text to Google Sheets, Airtable, Notion, a vector database, Slack, email, Make, Zapier, n8n or a webhook after each run.

Data source and legality

The Actor downloads only the image and PDF URLs you provide (or that are in the dataset you choose) and runs OCR locally with Tesseract. It respects robots.txt, sends a clear user agent and does not log in or bypass any protection. No image is sent to an external AI or OCR service.

You are responsible for having the right to process the images (copyright, personal data). Text in images can contain personal data, for example names on receipts or ID documents; process such images only with a legal basis and delete the results when you no longer need them.

FAQ

How accurate is it? On clean screenshots and printed documents Tesseract is usually above 90% confidence per word (the sample above: 91.4). Photos with perspective, low light, handwriting or decorative fonts are much harder; use the Sparse layout and a minimum confidence, or a vision-language model for those.

Does it read handwriting? Only neat block letters, and not reliably. Tesseract is built for printed text.

My PDF already has selectable text. Should I use this? No: the PDF Text Extractor reads the existing text layer faster, more exactly and for less money. Use OCR for scanned PDFs and PDFs that are images.

Which languages can I use together? Any of the 12, e.g. ["eng", "swe"] for Swedish documents with English terms. Each extra language makes OCR slower.

What does "Sparse" layout do? It looks for text anywhere in the image without assuming columns or paragraphs, which works better for posters, photos, memes and UI screenshots with scattered labels.

Multi-page TIFF or animated GIF? Only the first frame is read; pageCount shows how many frames the file has. Convert multi-page TIFFs to PDF to OCR every page.

Are images stored? No. Images are downloaded into the run's memory, read and discarded. Only the text is saved in the dataset.

Questions or bugs? Open an issue on the actor page. We reply within 24 hours. If the actor helps you, a rating on the Store page helps others find it.

Related actors

• PDF Text Extractor: text and Markdown from PDFs that have a text layer, with RAG chunks.
• Bulk Image & File Downloader: download all images from web pages or a dataset, then OCR them here via the dataset input.
• Website to Markdown: clean Markdown text of web pages for LLMs.

Changelog and support

• 1.0 (October 2026): first release. Tesseract OCR for image URLs, scanned PDF pages and dataset inputs; 12 languages; layout modes; confidence filter; optional line boxes.

Questions or feature requests? Open an issue on the actor page. We reply within 24 hours.