Image to Text OCR: Extract Text from Images (26 Languages)

Extract text from images, scans, screenshots and photos of documents. Paste image links and get back the text of each image, a confidence score and, if you want, every line and word with its position. Built on Tesseract 5 with smart preprocessing, so you don't have to clean up images first.

• ✅ 26 languages: English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Czech, Slovak, Swedish, Danish, Norwegian, Finnish, Turkish, Russian, Ukrainian, Romanian, Hungarian, Greek, Chinese (simplified and traditional), Japanese, Korean, Arabic, Hindi
• ✅ Auto-rotate: sideways and upside-down scans are turned upright before reading
• ✅ Small text works: screenshots and small crops are upscaled automatically (Tesseract reads best at about 300 DPI)
• ✅ All common formats: PNG, JPEG, WebP, BMP, GIF and multi-page TIFF (one result per page), transparent PNGs, 16-bit scans, phone photos with EXIF rotation
• ✅ Positions: optional line and word boxes (x, y, width, height) with a confidence each, for redaction, highlighting or form parsing
• ✅ Batches: paste many links, several images are read in parallel
• ✅ Fair billing: you pay per image read; broken links, non-images and failed images are free

Input example

{
    "imageUrls": ["https://raw.githubusercontent.com/tesseract-ocr/test/main/testing/phototest.tif"],
    "languages": ["eng"],
    "layout": "auto",
    "autoRotate": true,
    "outputDetail": "lines"
}

Field	What it does
imageUrls	Direct links to images (up to 40 MB and 60 megapixels each)
languages	Language codes such as eng, deu, chi_sim; several at once work
layout	auto for documents, sparse for screenshots and photos with scattered text, block or line for a single block or line
autoRotate	Turn sideways scans upright (default on)
outputDetail	text, lines (with boxes) or words (lines plus every word with its box)

Output example

Real output from a test run (lines shortened to the first two):

{
    "url": "https://raw.githubusercontent.com/tesseract-ocr/test/main/testing/phototest.tif",
    "fileName": "phototest.tif",
    "page": null,
    "width": 640,
    "height": 480,
    "rotation": 0,
    "languages": ["eng", "nld", "hin"],
    "text": "This is a lot of 12 point text to test the\nocr code and see if it works on all types\nof file format.\n\nThe quick brown dog jumped over the\nlazy fox. The quick brown dog jumped\nover the lazy fox. The quick brown dog\njumped over the lazy fox. The quick\nbrown dog jumped over the lazy fox.",
    "confidence": 95.3,
    "wordCount": 60,
    "charCount": 285,
    "lines": [
        { "text": "This is a lot of 12 point text to test the", "confidence": 95.4, "box": { "x": 36, "y": 92, "w": 544, "h": 30 } },
        { "text": "ocr code and see if it works on all types", "confidence": 95.3, "box": { "x": 36, "y": 126, "w": 582, "h": 31 } }
    ]
}

Line breaks follow the image; paragraphs are separated by a blank line. page is set for multi-page TIFFs and animated GIFs. rotation tells you by how many degrees the image was turned. A summary per link (ok or error with the reason) is saved as OUTPUT in the key-value store.

Pricing

You pay per image (per page for multi-page TIFFs). In our tests 4 images from 0.3 to 11 megapixels in 3 languages took 11 seconds with the default 1 GB memory. In longer runs Apify limits the CPU to the run's share of memory: 36 text-heavy images (12 megapixel document photos, screenshots and receipts) took about 9 minutes with 1 GB. More memory means more CPU, so long runs finish sooner; the price per image stays the same.

Paid Apify plans pay less: 10% off on Starter, 20% on Scale and 30% on Business and higher plans.

FAQ

Which languages should I pick? Only the ones in your images. Each extra language makes reading slower and can lower accuracy, for example by mixing up similar letters.

Does it read handwriting? Clear block letters often work; cursive handwriting does not. Tesseract is made for printed text.

My images are PDFs. Use our PDF to Markdown & Text Actor: it reads the text layer directly and only OCRs scanned pages.

What does the confidence mean? The average certainty of Tesseract per word from 0 to 100. Above 85 is usually clean text; below 60 the image is likely blurry, very small or handwritten.

More tools from Tidelane Data

• PDF to Markdown: PDFs to Markdown and text, with tables and OCR for scans
• Audio & Video to Text: speech to text transcription with Whisper
• AI Image Upscaler: upscale images 2x, 3x or 4x with Real-ESRGAN
• Bulk Image Downloader: all images from web pages, also as one ZIP

All our tools: apify.com/tidelane

---

Impressum

Levi Griesinger, Tidelane Data
Besigheimer Straße 18
74369, Löchgau, Deutschland
E-Mail: LeviGriesinger@gmail.com