Image to Text (OCR)
Extract text from images using AI-powered OCR. 100% private — runs entirely in your browser. Supports 18+ languages.
Upload any image and our browser-based OCR engine will extract all readable text instantly — no signup, no server uploads.
Powered by Tesseract.js WASM. Your images never leave your device.
Drop an image here or click to upload
PNG, JPG, WebP, BMP, GIF — Max 20 MB
How OCR Works
OCR (Optical Character Recognition) analyses pixel patterns in an image to identify letters, numbers, and symbols. The software breaks the image into small regions, compares each against a trained model of known characters, and outputs the best match as editable digital text.
This tool uses Tesseract.js, a WebAssembly port of Google's Tesseract OCR engine — the same technology that powers Google's document scanning. It runs entirely in your browser, so your images never leave your device.
Modern OCR handles printed text very well (95-99% accuracy with clean images). Handwriting recognition is improving but less reliable. For best results, use clear, high-contrast images with straight, well-lit text.
What to Expect: Accuracy by Input Type
| Input Type | Typical Accuracy | Tips |
|---|---|---|
| Printed document (scan) | 97-99% | High-res scan, clean background |
| Screenshot with text | 95-99% | Use original resolution, no scaling |
| Phone photo of document | 85-95% | Shoot straight-on, good lighting |
| Receipt or label | 80-90% | Small text is harder — get close |
| Text on complex background | 60-80% | Crop to just the text area first |
| Handwriting (neat print) | 50-70% | Block capitals work best |
| Cursive handwriting | 20-40% | Not reliable — type it manually |
Getting Better Results
Use 300+ DPI images. Higher resolution gives the engine more pixels to work with per character. Low-resolution screenshots of small text are the #1 cause of bad results.
Crop to just the text. Remove borders, images, and decorative elements. The less noise the engine has to filter, the better it focuses on the actual characters.
Select the right language. Each language has its own character model. English is loaded by default, but switching to the correct language before extraction dramatically improves accuracy for non-English text.
Watermarked or overlaid images. Text over photos, watermarks across documents, and coloured backgrounds behind text all confuse the recognition engine. Clean, high-contrast input gets clean output.
Heavily compressed JPGs. JPEG compression artefacts blur the edges of characters, especially at small sizes. If possible, use PNG or a high-quality JPG.
What to Do After Extraction
- Count words to check the extracted text length and reading time.
- Fix the case if the OCR misidentified capitalisation throughout the text.
- Clean whitespace to remove extra spaces and blank lines that OCR often introduces.
- Find and replace common OCR errors — like "l" misread as "1" or "0" misread as "O".
Related Tools
How to use this tool
Upload an image or paste from clipboard — drag-and-drop works too.
Select the language of the text in the image for best accuracy.
Click 'Extract Text' and copy the result to your clipboard.
Common uses
- Extract text from scanned documents and PDFs
- Digitise printed notes, receipts, and invoices
- Copy text from screenshots and images
- Convert handwritten notes to digital text
- Translate text in foreign-language images
- Extract data from charts and infographics
Share this tool