Fonts have long provide a way to present text without having to worry about differing resolutions and point sizes, but on the web you’ll still come across the odd piece of text that, for one reason or another, is displayed in raster, rather than vector form. Interacting with said content just became much easier, thanks to Project Naptha.
Project Naptha is an extension for Chrome that can detect text in images using the “Stroke Width Trasform”, a character recognition algorithm developed by Microsoft Research. Its strength lies in its ability to pick out text in a “language-agnostic manner”.
With the extension installed, I attempted to use it to grab text from images in Chrome’s Web Store… without success. It appears Project Naptha doesn’t work on WebP, Google’s JPEG successor. Fortunately, the rest of the internet has yet to jump on this particular train and once I loaded a few sites using more traditional formats, the extension began to work.
There’s a short delay as it processes the text, but you’ll know when it’s done when the familiar crossbar cursor appears. You can then drag and highlight text and perform a number of operations from copying and pasting to “erasing”, which will magically remove the selected letters (most of the time).
When there’s a clear contrast between the text and background the extension works well, however, it understandably gets confused if there’s a lot going on. That said, it’s an impressive extension and while its uses are niche right now, it’s one I’ll definitely be keeping installed, just in case.