Nowadays, most of the content present on social media visuals is photo or video. What many may not know is that, for people who are blind or have low vision, navigating these can be a problem. This is because, when passing through the timeline, VoiceOver (or any other software used for reading the screen) reads everything that is text, but leaves out photos and videos.
For this reason, on March 29, Twitter announced the possibility of adding descriptions to the photos which will certainly help a lot. Facebook, however, went further and created a tool that helps to ?read? the images automatically.
The tool, called ?Automatic Alternative Text? (?Automatic Alternative Text?) developed by the Facebook accessibility team, uses object recognition technology to automatically generate descriptions for photos published on the network. In this way, when passing through a photo, the reader can describe ?this image can contain: two people, smiling, sunglasses, nature, water? and things like that. This technology based on Neural Networks with billions of parameters and trained to recognize millions of examples.
Descriptions are loose words that look like tags which automatically appear in the photos to ?explain? to the user what is in the photo. At first, images and words are recognized in transportation ("car", "boat", "motorcycle", etc.), nature ("mountain", "wave", "sun", etc.), sports ("swimming" , ?Stadium?, ?sneakers?, etc.), food (?ice cream?, ?sushi?, etc.) and some words that describe the appearance (?baby?, ?smiling?, ?selfie?, etc.).
In the video below, you can get a sense of how a blind person uses VoiceOver and how this new feature works:
The novelty was released yesterday (April 4) for iOS and, for now, only in English and in the following countries: United States, United Kingdom, Canada, Australia and New Zealand. It will soon arrive on Android and we are hoping that it will be available soon in Brazil too. ?