Extract Text from Image · Last updated: August 18, 2026
We do not collect any personally identifiable information, browsing history, page content, screenshots, images, or recognized text.
We collect anonymous usage statistics: when the extension is installed, used (one "active today" ping per day) or uninstalled, it sends a small event to our server containing a random identifier (a UUID generated inside the extension, not linked to you or your Google account), the extension version, and your browser interface language.
The same kind of event is also sent for basic interaction steps and failures, so that we can tell a working feature from a broken one: opening the popup, starting a text extraction, finishing one (including whether it returned no text), a recognition error together with a short technical reason code (for example, a model file failing to download), changing the recognition language, whether the toolbar icon is pinned, and opening a page where screen selection is not permitted by Chrome. These events never contain your images, screenshots, recognized text, page addresses or any part of what you extract — only the type of step and the extension's own version and language settings.
Our server does not store IP addresses or user-agent strings. The IP address of a request is used once, in memory, to derive a country code (for example, "DE") for aggregate statistics, and is then discarded without being written anywhere.
If you choose to rate the extension and leave a comment, or select an uninstall reason on the page shown after removal, the text you enter is sent to our server together with the same random identifier and extension version. This is entirely optional and contains no email address or account information.
When you click the extension icon and select a region of a webpage, the extension captures a screenshot of that region using Chrome's built-in API. You can also choose to upload an image file from your device instead, either from the popup's "Upload File" tab or from the "Upload image file" option shown after clicking "Clear" in the results panel. Either way, the image is processed entirely within your browser using a local PaddleOCR machine learning model (WebAssembly) — no image, text, or metadata is sent to any external server. The recognized text appears in a panel inside your browser tab, where you can copy it or download it as a local .txt file.
On first use, the extension downloads OCR model files (~15 MB total) from a public GitHub repository:
This is a standard HTTPS download, identical to loading any public webpage. GitHub may log the request as part of standard server operations. After download, model files are cached locally — no further network requests are made for OCR processing.
The extension stores your selected language preference using Chrome's storage.local API. This data stays on your device and is never transmitted anywhere.
The extension does not use any third-party analytics, advertising networks, or tracking pixels. Anonymous usage statistics and optional feedback are sent to our own self-hosted server (devexthub.com) and are never shared with or sold to anyone. The only other external request is the one-time download of OCR models from GitHub (described above).
To remove all data associated with this extension, simply uninstall it from Chrome. This clears all stored preferences and cached model files. Because usage statistics are tied only to a random identifier, they cannot be linked back to you; if you want the events for your random ID deleted from our server, contact us via the Chrome Web Store listing.
If we change this policy in the future, the "Last updated" date above will be modified. Material changes will be communicated via the Chrome Web Store listing.
Questions about privacy? Open an issue on the project's GitHub repository or contact the developer through the Chrome Web Store listing.