Free
Image Converter
Convert images between JPG, PNG, WEBP, AVIF, GIF, ICO, PDF and more, and resize, crop or rotate them in bulk.
Image to Text Converter (OCR)
Pull the text out of screenshots, photos and scanned pages in English, Hindi, Marathi and 22 more languages. The Tesseract OCR engine runs in your browser, so your images never leave your device.
Add images
Drop images here
or choose files · or paste a screenshot with Ctrl+V
PNG · JPG · WEBP · BMP · GIF · PDF
Drag on the image to read only part of it. The preview shows the clean-up settings below.
Mixed text:
Image clean-up and layout
Pixels lighter than this become white, darker ones black. Raise it if thin letters disappear.
Getting ready…
Your images are read on this device and never uploaded. The OCR engine (Tesseract.js) and the language files are downloaded from the jsDelivr CDN.
Extracted text
The text from your images will appear here, ready to edit, copy or download. Add an image or try the example.
About the tool
How the Image to Text Converter works
The Image to Text Converter reads the words in a picture and turns them into text you can edit, copy and search. Drop in a screenshot, a photo of a document, a scanned page or a PDF, or simply paste a screenshot with Ctrl+V. You can add many images at once and read them in one go, then copy or download each text on its own or all of them together in one .txt file.
It uses Tesseract, the well-known open-source OCR engine, compiled to WebAssembly so it runs inside your browser. Tesseract recognises text with an LSTM neural network, a type of machine-learning model, and this tool uses its compact "best_int" models. The first time you use it, your browser downloads the engine (about 1.5 MB) and the data for the languages you choose (about 0.7 to 3 MB each, for example 2.8 MB for English and 1.3 MB for Hindi) from the jsDelivr CDN. These files are checked against fixed checksums and saved in your browser, so later visits start straight away. Your images are processed on your own device and never uploaded.
Clear, printed text in a good-quality image usually comes out almost perfectly, but OCR is never 100% accurate. Each result shows Tesseract's confidence score, and the word-box view marks the words it was unsure about so you can check them quickly. If the result is poor, try the clean-up options (Invert colours helps with light text on a dark background), choose the right language, rotate the image the right way up or drag a box around just the text you need. Handwriting, very small or blurry text and decorative fonts are hard for any OCR engine.
Frequently asked questions
Are my images uploaded to a server?
No. The OCR engine runs in your browser, so your images and the text found in them stay on your device. The only downloads are the Tesseract engine and the language files, which come from the jsDelivr CDN the first time you use a language and are then saved in your browser.
Does it use AI?
Yes, in a narrow sense. Tesseract recognises text with an LSTM neural network, a machine-learning model trained to read printed characters. It runs entirely on your device. It does not use a chatbot or a large language model, and it does not rewrite or correct your text.
How much does it download, and why is the first run slower?
The engine is about 1.5 MB and each language is about 0.7 to 3 MB (English 2.8 MB, Hindi 1.3 MB, Marathi 1.9 MB). They are saved in your browser after the first run, so later runs only need a second or two per image. You can delete the saved files with the link under the Extract button.
Can it read Hindi, Marathi or a mix of English and Hindi?
Yes. Pick Hindi or Marathi as the language, or use a mixed pair such as English + Hindi when one image contains both scripts. Choosing only the languages that are actually in the image gives the best results and the smallest download.
Why are some words wrong, and how can I improve the result?
OCR works best on sharp, straight, printed text. Turn on Show word boxes to see which words had low confidence. Then try Invert colours for light text on a dark background, Black and white for faint scans, rotating the image the right way up, a different page layout, or dragging a box around just the text you need. Handwriting is rarely read well.
Can I extract text from a PDF?
Yes. Each page of the PDF (up to 30 pages per file) is turned into an image in your browser and read like any other picture. This is meant for scanned PDFs. If your PDF already has selectable text, copying it directly from a PDF reader is faster and exact.
More free AI tools
- Image
- ImageFree
Image Compressor
Shrink JPG, PNG, WEBP and AVIF images in bulk, set a target file size and compare before and after.
- ImageFree
Image Resizer & Cropper
Resize images to exact pixels or a percentage, crop to social media sizes, rotate and flip, then save as JPG, PNG or WEBP.