Facebook improves AI photo descriptions for visually impaired

New technology helps Facebook offer labels.

Social Media Images

Facebook has been using AI to describe photos for people with visual problems, but it is reinforcing its efforts in 2021. The social media giant has detailed a new version of the automatic alternative text (AAT) That promises much more information.

instead of relying on highly supervised AI learning, Facebook is now using low monitoring based on “billions” of Instagram photos and hashtags.

Facebook and Google reportedly reached a settlement that reduced advertising competition

The method allows Facebook to expand more than 100 descriptions of concepts to include more than 1.200, such as different types of food and national monuments. It’s also more culturally inclusive-you can recognize weddings that don’t involve white wedding dresses, for example.

a new object detection system can also recognize where people are in the frame as well as the number of people in the scene.


And while you will normally get a simple summary of the contents of a photo, you will have the choice of a detailed description describing the position, size, and nature of the objects.

It still offers an imperfect description-Facebook pointed out that it still uses “maybe” and errs on the side of excluding concepts that it can not recognize solidly.

Facebook’s Libra foreign money to release subsequent 12 months in restrained layout

However, it is a big leap for accessibility on the social network, and it could ensure that more people with visual problems at least have the essence of the photos in their feed.

Engadget / TechConflict.Com

Contact Us