Facebook’s developers have created a new AI, called Rosetta that can read text from billions of images every day. So now facebook can read and make sense all of those memes and images we share carelessly.
Facebook has ramped up its efforts to better scrutinize content post the US-election controversy and ‘Rosetta’ is the latest attempt in that regard. Rosetta can match the text with its image and identify potentially offensive content or hate speech. Mark Zuckerberg has called it a “weapon” against fighting hate speech.
The technology will be able to read menus and road signs in images as well as words on clothing etc. The process involves scanning images for text and then using text recognition and image processing techniques to identify what the text says. After transcribing the image to text, the system then interprets the text in the context of the image to make sense of what it says.
Facebook’s developers have said that the technology can recognize texts in a wide variety of languages which will increase its utility in the long term as well. Facebook does not only plan to utilize this technology to simplify the process of searching images or screen out content that violates facebook’s hate speech policy. It also plans to make use of it to personalize content. For future, Facebook hopes to expand the AI to video content as well, however, that will be much more laborious and complex project.