Image to Text (OCR)

OCR runs entirely in your browser via WebAssembly. Images never leave your device.

Drop images or scanned PDFs here, or click to select. You can also paste a screenshot with Ctrl+V.

Download size: English ~10MB, Chinese ~19MB. Cached after first use.

No files yet

  1. Drop images or a scanned PDF, or paste a screenshot with Ctrl+V.
  2. Pick the recognition language (English / Simplified Chinese).
  3. Click "Recognize Text" and wait for the progress bar.
  4. Copy or download the recognized text.

Help

FAQ

Where does recognition happen? Are my images uploaded?Show
Recognition runs entirely in your browser via WebAssembly — images never leave your device. The only network use is a one-time download of the engine and language data (~10MB English, ~19MB Chinese), cached by the browser afterwards.
Which languages are supported?Show
Currently English and Simplified Chinese. The Chinese language pack is larger (~19MB) and takes longer to load the first time.
Can it recognize scanned PDFs?Show
Yes. Each PDF page is rendered to an image with pdf.js and then recognized — works for scanned documents and image-only PDFs. Many-page PDFs take longer.
The results are poor. What can I do?Show
OCR works best on sharp, upright, high-contrast text. Blur, rotation, stylized fonts, and low-resolution screenshots all hurt accuracy. Capture screenshots where the text fills the frame.
What does the "Keep line breaks" toggle do?Show
Off (default) merges the result into flowing paragraphs, good for reading. On preserves the original line structure, good for code screenshots, poems, or table-like content.

How to convert an image to text

  1. Drop images or a scanned PDF, or paste a screenshot with Ctrl+V.
  2. Pick the recognition language (English / Simplified Chinese).
  3. Click "Recognize Text" and wait for the progress bar.
  4. Copy or download the recognized text.

Privacy

Full guide

What this tool does

This page extracts text from images using OCR (optical character recognition) and turns it into editable, copyable text.

Common uses:

  • Extract text from screenshots (chat logs, error messages, code snippets)
  • Recognize scanned PDFs, contracts, and paper documents
  • Pull text from course video frames or photos
  • Batch-process several images and copy everything at once

Supported inputs

  • Images: JPG / PNG / WebP / BMP and other common formats
  • Scanned PDFs: each page is rasterized and recognized one by one
  • Clipboard paste: press Ctrl+V to paste a screenshot directly — no need to save a file first

Getting better accuracy

  • Use sharp, high-resolution images — OCR is very sensitive to blur and low resolution
  • Keep text upright (the tool does not auto-rotate)
  • Frame screenshots so the text dominates, avoiding noisy backgrounds
  • Dark-background screenshots usually recognize worse than light-background dark text

Privacy

The recognition engine (Tesseract.js, the WebAssembly build of Tesseract) runs entirely in your browser. Neither images nor results are uploaded. First use downloads the engine and language pack from a CDN; afterwards the browser caches them, and recognition keeps working on an already-open page even offline.

Related tools