In 2016, Facebook introduced the Automatic Alt Text (AAT) system for blind people for its app and Instagram, with the help of which people can easily know about any image. This system helps in generating the description of any image. Now the company has announced that an automatic alt text system can now know 10 times more about any object. With this, the description will also be in more detail and will be able to identify AAT activity, landmarks, type of animals, etc. in a better way.
The company has stated in its blog post that the AAT improvised can recognize more than 1,200 concepts. It is 10 times better than its original version. Facebook also said that AAT will also provide information about the position and size of the element present in the picture. As an example, instead of saying ‘Facebook is probably five people in this image’, this AI will specify that there are two people in the center of this photo and three people are spread towards the fringe.
AAT by default will provide users with compact descriptions for all photos. With this, users will get a detailed description of the photo, which can develop more interest in people. When the user selects this option, a panel will come in which more information about the content of the photo will be found.
Facebook has said that the company has consulted users who depend on screen readers and found that they want more information about images coming from friends and family. In such a situation, it will provide a detailed description of the ATT system image, which will also include positional information.