Automatic Alternative Text is a new tool that Facebook is implementing for people who cannot see are not excluded from the conversation on social networks, especially the enormous volume of images that all published there.

There are 39 million blind people in the world according to WHO

There are several tools for blind surf the Internet: one of the most common is to use software that looks at the code behind the page and reads them aloud texts and links (or even transforms them into braille).It is a bit cumbersome, slow and sometimes confusing, especially if the code on that page is not ready.

In social networks, the images that our friends or acquaintances share are a huge part of what we consume every day and although the software can alert you that there is an image, has no way to translate for its content into words … Until now.

How Facebook will make the blind to “see” the images?

Facebook is wearing a new artificial intelligence system that can recognize more than 80 things and ways that our images usually have in common, including:

Facial and body expressions like smile or hug – for example, a voice would say “two, smiles.”
Means of transport such as cars, bicycles, buses or trains.
Places like the beach, outdoors, mountains, if it is sunny, if there is grass and more.
Physical activities, from sports, “walking”, “playing football”.
Appearance: sunglasses, beard, selfie (of course).
Food like pizza, birthday cake, sushi, coffee, to make you salivate.

The project uses technology similar to Reconnaissance faces with which Facebook already has and was the initiative of Engineer Matt King who works there and was gradually losing his sight by disease.

Last month Twitter added a feature where users manually put tags to your own images. This is more direct, automatic (digital, if you want), and Facebook believes that if the rest of us can see, this software is not violating anyone’s privacy. Face recognition has already provoked comments of this type.

Share this on WhatsApp