How-to · OCR
How to copy text from a screenshot in Chrome
The short answer
Capture the region of the page that contains the text — or paste an existing screenshot — and let a local OCR engine turn it into selectable text. Everything is recognized in your browser: no upload, no account, works offline.
- 1
Click the Image OCR icon on the page with the text you want.
- 2
Choose capture and drag a box around the text region — or paste an existing screenshot into the workspace with Ctrl+V.
- 3
Let the bundled OCR engine recognize the region locally; nothing is uploaded.
- 4
Proofread the editable result against the original image.
- 5
Click "Copy all" and paste the text wherever you need it.
Some text just won't let you select it: an error dialog, text baked into an image, a video frame, a PDF rendered as pictures, a chart label. The usual workaround — retyping it by hand, or uploading a screenshot to an "image to text" website — is either slow or means handing your screen contents to a server. In Chrome, an OCR extension can do the whole job locally.
Capture the region with the text
Click the Image OCR icon and choose the capture option. The extension takes a screenshot of the current tab (using Chrome's activeTab permission, so it only sees the page you invoke it on) and lays a framing overlay on top. Drag a box around the text you want — just the dialog, just the caption — and only that region goes to the recognizer. Tighter selections mean faster, cleaner results.
Or paste a screenshot you already have
If the screenshot already exists — from your phone, a colleague's message, your OS screenshot tool — open the workspace instead and paste it with Ctrl+V or drop the file in. PNG, JPG, WebP and BMP all work. Capture and file input feed the same local pipeline.
The text is recognized on your device
Recognition runs on a Tesseract engine compiled to WebAssembly and bundled inside the extension, together with its language data. Nothing is fetched at runtime and nothing is sent out — your screenshot never leaves the browser, and the whole thing works with Wi-Fi off. The recognized text appears next to the original image in an editable box; click the original to compare when a word looks off, fix any misread characters in place, then hit Copy all. There is no account and no per-capture limit.
If a page can't be captured
Chrome blocks extensions from injecting into browser-internal pages (chrome:// settings, the Web Store and similar), so the capture overlay can't run there — the popup will tell you so. The workaround is the one you'd expect: take a normal OS screenshot of that page and paste it into the workspace. Same engine, same result.
Mind the recognition language
If the screenshot is not in the language the recognizer expects, the output turns to soup. Switch the recognition language in the popup before capturing — the switch is instant because the language data ships inside the extension. For images that live in files rather than on screen, see the companion guide on extracting text from an image in Chrome.