Image to text: extract text from a photo or screenshot
Copy the text out of a photo or screenshot with on-device OCR for English and Arabic, with clean-up for faint or small text.
Loading tool…
Retyping text from a screenshot, a photo of a notice or a scanned letter is slow and error-prone. This tool runs optical character recognition (OCR) with Tesseract, the open-source engine originally developed at HP and maintained by Google for years, compiled to WebAssembly so it works inside your browser. The image is never uploaded: the engine and its language data are downloaded from this site the first time you click Extract text, then the browser reuses them.
You can read English, Arabic, or a mix of both in one image. Before reading, the image can be converted to grayscale, have its contrast stretched (useful for faint text or photos of paper) and be enlarged when it is small, because Tesseract needs characters roughly 20 pixels tall to read them reliably. Screenshots of apps and websites usually come out almost perfect; phone photos work well when the page is flat, sharp and evenly lit.
After reading, the tool shows Tesseract's average confidence, so you know how carefully to proofread. OCR is not perfect: handwriting, decorative fonts, very low resolution, tables and text on busy backgrounds are where mistakes appear. For multi-page scanned PDFs, use OCR PDF instead, which also makes the PDF searchable.
How to use it
- Drop a photo or screenshot (JPG, PNG, WebP, HEIC and more) or click to choose one.
- Pick the language of the text: English, Arabic, or English + Arabic for mixed text.
- Leave the clean-up options on (grayscale, contrast, enlarge small images) unless the result gets worse, then click Extract text.
- Check the text against the image, edit anything that was misread, then copy it or download it as a .txt file.
Frequently asked questions
Is my image uploaded?
No. The OCR engine (tesseract.js and its WebAssembly core) and the language files are downloaded from this site to your browser, and the image is processed on your device. You can disconnect from the internet after the engine has loaded and it still works.
How accurate is it?
For clean screenshots and printed text it is usually very accurate. Accuracy drops with blur, perspective, low resolution, unusual fonts, handwriting and text over pictures. The confidence figure shown after each run is Tesseract's own estimate; below about 65% expect several mistakes.
Does it read Arabic?
Yes. Choose Arabic, or English + Arabic for mixed text. Printed Arabic in common fonts (such as Naskh-style fonts in documents and apps) reads well; diacritics (tashkeel) are often dropped or misread, and handwriting or calligraphy is rarely recognised.
How big is the download?
About 7 MB for English, 6 MB for Arabic and 8.5 MB for both, downloaded once. That includes the recognition engine (about 3.9 MB) and the trained data for each language.
Can it keep the layout, columns or tables?
It keeps line breaks and spacing between words, but it does not rebuild tables or columns. For documents where layout matters, OCR PDF produces a searchable PDF that keeps the original page image.
Related tools
- OCR a scanned PDFMake scanned PDFs searchable: Tesseract reads English and Arabic text and adds an invisible text layer.
- Extract text from a PDFExtract the text of a PDF in reading order and download it as a .txt file or copy it.
- Crop an imageCrop to 1:1, 4:5, 16:9, 9:16 or any size by dragging or typing pixels; circle crop to transparent PNG.
- Upscale an image 2x, 3x or 4xEnlarge an image 2x, 3x or 4x (or to a width) with Lanczos-3 resampling and optional sharpening.
- Convert HEIC to JPGConvert iPhone HEIC/HEIF photos to JPG that opens anywhere, without uploading.
- Word counter and character counterCount words, characters (with and without spaces), sentences and paragraphs, estimate reading and speaking time, and see keyword density.