The new artificial intelligence of Instagram informs users with vision problems of what is on the screen

The accessibility measures that modern smartphones incorporate are one of the major recent contributions of technology to people with disabilities. The big tech companies responsible for the main operating systems, Apple and Android, as well as the other players involved, have taken the issue very seriously for some time. Now, Instagram adds artificial intelligence (especially, machine learning) to detect what is in the downloaded photos and say it out loud to people with vision problems.

Instagram calls this function automatic alternative text, and allows you to listen to the descriptions in the most used parts of the application: Stream, Explorer and Profile. In machine learning, according to the company, it is used object recognition technology, with a model formed.

In addition to automatic detection, the manual alternative text will also be a reality on Instagram. Now, when uploading photos, users can write a description of the added images so that users with vision problems can learn more.

How does image recognition work in other social networks?

Facebook announced in 2016 the arrival of photo description function by artificial intelligence, but only for its main social network. For him, strange that it took more than two years to bring something that is a great help for millions of users on Instagram. Facebook also allows you to add text to users

Accessibility has also been important for Twitter, which since 2016 also allows users to upload photos. add textual descriptions of the content. The sanctions are that, the function, for the moment seems far from what Instagram and Facebook have implemented.

I am a blind user of Twitter. We are many there. Increase your ability to reach us and help us interact with your images. It's very simple and makes a huge difference in our Twitter experience, allowing us to see your images in our own way. Thanks for the description pic.twitter.com/hCsjoFdmev

– Rob Long (@NetworkLong) January 3, 2018

The explanation is clear Twitter does not have an automatic object recognition model, but this requires the user to write the description, which, although some users have explained and requested, is not usual.