Facebook AI technology describes photos to blind users
Dubbed automatic alternative text, the new feature generates photo descriptions for visually impaired users who rely on a screen reader.
Prior to adding the new feature, Facebook users of screen readers would hear the name of the individual who posted the image, followed by the word “photo” when they would scroll past an image in their news feed.
The new feature now adds a basic description of the photo such as, “image may contain four people, smiling, outside.”
The feature uses the social network’s visual recognition engine, an artificial intelligence-powered technology that processes all photos and videos uploaded to the website.
To date, the engine is able to recognize several items including car, mountain, tree, sunset, basketball court, ice cream and pizza. The feature then includes these objects in its captions for the visually impaired.
At launch, the automatic alt text feature offers about 100 concepts, which has a successful identification accuracy of 80 to 90 percent.
Facebook is slowly adding more items, but is focusing more on precision rather than quantity in order to ensure that the tool does not mislead blind users.
First introduced as a prototype back in November, automatic alt text will also be expanded to Android devices and desktop users in the future. Though the initial version of the feature is in English, Facebook will systematically add other languages in the future.
Meanwhile, there has been some debate as to whether or not all websites should have to adhere to the same standards of being accessible to everyone, including those with disabilities.
When asked if all social media users should be required to add these tags to images, one expert said that such a requirement would cause unnecessary burden on social media users.
“In a perfect world all users of social media would take a moment to provide a brief description of an image being posted,” said Ryan Jones, a certified vision rehabilitation therapist from Freedom Scientific’s services and training department. “This is not, and probably never will be, the reality though as most people either don’t have the time, or don’t understand the importance of providing a text description of an image for those of us who cannot see it.”