Image to Text OCR – Extract Text from Images Online (Free)
Free online OCR tool to extract text from images (PNG, JPG, WebP, BMP). Supports English, Vietnamese, Chinese, Japanese, Korean, French, Spanish, German, Russian. Powered by Tesseract.js — runs locally in your browser, no upload required.
Example Output
Tesseract OCR output preserves line breaks and Unicode diacritics — Vietnamese accents like ấ, ử, ợ come through correctly.
Plain UTF-8 text: full page contents transcribed line by line
What is Image to Text (OCR)?
Extract text from any image directly in your browser — screenshots, photos of receipts, scanned book pages, posters, presentation slides. The tool uses Tesseract.js, the de-facto open-source OCR engine, running in WebAssembly so nothing is uploaded. Supports 10+ built-in languages including English, Vietnamese (with full diacritic support), Chinese, Japanese, Korean, and major European languages.
Why use this tool?
- Touch-friendly UI, fine on phones for on-the-go edits
- No registration, account, or installation required
- No API keys to manage and no rate limits to monitor
- Your original file is never uploaded — sensitive screenshots stay on your device
- Output preserves color profile and metadata except when stripping is explicitly requested
How to use
- Pick the OCR language (or English + Vietnamese for mixed docs)
- Drag-and-drop or upload an image (PNG, JPG, WebP, BMP)
- Click Run OCR — first run downloads training data
- Copy the extracted text or download as .txt
Examples
Screenshot to editable text
A screenshot of a forum post becomes editable plain text you can paste into a doc — no manual retyping.
Receipt digitisation
A photo of a Vietnamese restaurant receipt is OCR'd with the `vie` language pack — every diacritic preserved.
Slide-deck text extraction
Screenshots from a teammate's slide deck become searchable text for grep/notes — useful when only PNGs are shared.
Common use cases
- Digitising printed documents (books, receipts, contracts)
- Extracting text from screenshots without retyping
- Building searchable archives of image-based content
- Pulling quotes from poster / signage photos
- Preparing image text for LLM ingestion or translation
Troubleshooting
- OCR misses some Vietnamese diacritics
- Pick the `vie` language pack (or `eng+vie` for mixed-language docs). The default `eng` model doesn't know Vietnamese accent rules.
- Output is gibberish for a clean photo
- The image is rotated or the text is too small. Rotate to upright orientation and upscale to at least 300 DPI / 1000 px wide before OCR.
- First OCR takes 20+ seconds
- Training data is downloading on first use (~5-15 MB per language). Subsequent runs use the cached model and start instantly.
Frequently Asked Questions
Powered by Tesseract.js (the same engine Google open-sourced from Tesseract). Accuracy is excellent for clean, high-resolution screenshots and printed text. Accuracy drops on handwriting, blurry photos, or text smaller than 12 pixels tall.
Try these related tools
Explore more Image Tools
Discover other free, privacy-first tools in Image Tools.