Facebook on Tuesday rolled out an update to its that improves the detail of photo descriptions for blind and visually impaired users. The tool, which , uses computer vision technology to automatically create descriptions. Updates are available when using a screen reader, an assistive technology that converts text to speech.
AAT can now recognize over 10 times more objects and concepts in photos on Instagram and parent company Facebook since the feature was introduced, bringing the total to over 1,200. This means more photos will have descriptions, Facebook says.
The company also expanded existing object and concept categories and added new ones. Now AAT can recognize landmarks, activities and types of animals, which allows for more detailed descriptions. An example of this could be, "May be a selfie of 2 people, outdoors, the Leaning Tower of Pisa." On Facebook, AAT can also offer additional details like the position of objects in a photo (top, middle, center and so on), as well as their relative size (primary, secondary). To access these detailed image descriptions, long-press on a photo in Android and use a "custom action" on a photo in iOS.
AAT is available for Facebook photos in News Feed, profiles and groups. On Instagram, it's available for photos in Feed, Explore and Profile. Because alt text descriptions are phrased simply, they can be translated into 45 languages, Facebook says.
Facebook joins a list of tech giants including Apple, Google and Microsoft that are placing a stronger emphasis on accessibility and inclusive design. The social network has launched a handful of accessibility features lately, including scalable font sizes and contextual headings. Last year, Instagram added automatic captions to IGTV.