Facebook has updated its automatic alternative text (ATT) feature that was first introduced in 2016, to be more efficient. The feature will now be able to recognize objects and generate captions or descriptions on photos better. This feature will also cater to the needs of Facebook users, who are visually impaired or blind.
Facebook’s automatic alternative text feature was awarded the Helen Keller Achievement Award from the American Foundation for the blind back in 2018. There are many users, who are blind or visually impaired (BVI), and now it will be easier for them to have a better understanding of the Facebook images with the help of a screen reader that will describe the contents of the images using a synthetic voice.
“Taken together, these advancements help users who are blind or visually impaired better understand what’s in photos posted by their family and friends — and in their own photos — by providing more (and more detailed) information” according to the Facebook blog post.
The ATT feature of Facebook integrates multiple technological advances that will enhance the user’s photo experience. The ATT will detect and identify the various concepts in the photo 10 times more efficiently, which means fewer photos without description. There will be more detailed descriptions as the feature will be able to identify and detect activities, landmarks, types of animals, people, and many more. The Facebook blog post had cited an example to better explain the feature, “Maybe a selfie of 2 people, outdoors, the Leaning Tower of Pisa.”
Facebook claims that this is the first achievement that made it possible to integrate information about the positional location and the relative sizes of the elements and objects in the photos. The feature will be able to specify how many people are there in a photo and precisely also locate the individuals who are scattered in the background of the photo-detecting the focus or the center in the photo. Not only people but it will also identify landscapes and the main object in the focus of the landscape photo.
According to the Facebook blog post, the ATT has been upgraded to detect over 1200 concepts and the screen reader users say that the accuracy is ‘paramount’. “We want to give our users who are blind or visually impaired as much information as possible about a photo’s contents — but only correct information”, the company said.