April 7, 2016
With infinite streams of visual content dominating the majority of social platforms, we are all more than aware of the highly visual nature of online communications. However, this can often leave the visually impaired marginalised from the online experience.
Facebook wants to change this. In an effort to help the blind community, the social media giant has launched a tool that will help the visually impaired to enjoy photos shared within the platform.
The feature, now available to all, is trained to recognise and identify a variety of objects within photos. Each image is analysed and automatically translated by Facebook’s AI into a rich audio description of the image – also known as “Automatic alternative text”, or “automatic alt text”.
Previously, if a blind user came across an image whilst navigating their newsfeed via a screen reader, the description would be minimal, merely reading the text within the update, the name of the person sharing that update, simply followed by “photo”.
With automatic alternative text the user may now hear:
“Image may contain: two people, smiling, sunglasses, sky, outdoor, water”.
If this image description is read against the caption: “We finally made it!”, the user is able to obtain a much richer understanding of the story.
The rationale behind using the phrase “image may contain”
Facebook cannot guarantee 100% accuracy of the audio description. If Facebook isn’t 80% certain of how to tag an image, it will not suggest a description.
At present, the feature only exists in iOS, and is available for English language users within the US, UK, Canada, Australia, and New Zealand.
If you would like to learn more about Facebook’s automatic alternative text, check out the video below: