OCR + Translate — Crivella

What is this?
A browser extension for running in-place translation of images in an active tab.

How does this work?
This is not a standalone tool!!!
The extension, when activated on a tab, will grab all existing and new images and send them to a backend server, that will perform the OCR and translation.
The server will reply with a list of textboxes, that the extension will overlay on top of the image

The idea behind this project, is to have a self-hosted instance of the backend server, which, by using pre-downloaded or self-generated/fine-tuned models is capable of running without any internet connection.

Why do you need to access your data on all the websites you visit?
This permission is required to allow the extension to inject its content script on all pages.
This is necessary to be able to grab all images and send them to the server for processing.

Documentation links

• Addon usage documentation
• Server readme
• Server detailed documentation

Privacy policy
Privacy policy
This tool does not collect any data from the user.
Depending on the plugin used , part of the box/ocr/tsl pipeline could be ran on outside services (This will be highlighted in the plugin description)

TLDR

• Install the extension
• Download and run the server
• Open the extension popup menu and use the "Plugins" section to install plugins in the server. Different plugins will enable different type of models.
• Select the source and destination language in the first form and hit submit
• Select the BOX/OCR/TSL models and click submit (could take some time especially the first time if the server has to download them)
• Use the page action (icon in the URL box) to activate the extension on a page