ocrs browser extension

The Ocrs browser extension allows you to copy text from images, videos, PDFs or any other content displayed in a browser tab.

It has currently only been tested in Chrome.

Building the extension

1. First, build the WebAssembly OCR library. In the root directory of this repository run:

   make wasm

2. Download pre-trained models. The easiest way to do this is to run the ocrs CLI tool, which will download models from the preferred location, and then copy them from the cache directory to <repo_root>/models/ocr. In the root of the repository run:

   cargo run -r -p ocrs-cli test-image.jpeg
   mkdir -p models/ocr
   cp ~/.cache/ocrs/text-detection.rten ~/.cache/ocrs/text-recognition.rten models/ocr

   Where test-image.jpeg can be any image you have available.

3. Navigate to this directory and build the browser extension:

   cd ocrs-extension
   npm install
   make build

4. In Chrome, go to chrome://extensions and select "Load unpacked extension", then select the ocrs-extension directory.

Using the extension

1. After installing the extension, click the puzzle piece icon in Chrome's toolbar and click the pin icon next to Ocrs to add it to the toolbar.

2. On any browser tab, click the Ocrs logo in the toolbar to take a screenshot of the current tab and highlight selectable text.

3. Click anywhere outside of a text region or press Escape to close the OCR overlay.