Facebook added automatic replacement text function

Today, the user experience has become the strategic focus of major companies. In order to improve the social networking user experience of the visually impaired, the social applications Twitter and Facebook have introduced new features one after another, so that visually impaired users can also "see" the pictures.

According to foreign media reports, the Facebook application update for iOS today added a new feature that allows visually impaired users to "see" photos.

It is reported that iOS version of the Facebook application can use artificial intelligence technology to automatically generate a description of each photo. This description is called "Automatic Alternative Text".

When users use screen readers on iOS, they can hear explanations about the list of items on the photo, such as "Photos may include three people, they are smiling, they are outdoors." If you do not use a screen reader The user can only hear a brief description about "a new photo was released".

It is reported that "automatically replace text" is a standard HTML attribute designed to enhance the capture of screen reading tools. "Automatically replace text" is called the artificial intelligence behind this function using Facebook's target recognition technology to identify things on the photo.

Jeff Wieland, head of the Facebook accessibility department, points out that "the most frequently used description of photos will be added to the tool to highlight the narrative focus and accuracy of the photos."

Currently, the automatic text replacement function only supports English and is suitable for iOS users in the United States, United Kingdom, Canada, Australia, and New Zealand. It can only recognize basic items such as cars, boats, basketball, baseball, ice cream, sushi, beard, and glasses in the picture. . However, Facebook said that the company will soon extend this feature to more platforms, languages, and countries.

Facebook said that it took 10 months to develop this feature to the current state, using a neural network that simulates the human brain to learn how to identify pictures and use numerous examples to train the recognition network, hoping to complete the inclusion of pictures. A more complete description of the identities of Chinese people, etc., expands the types of devices and languages ​​supported by the tool.

It is worth mentioning that, a few days ago, iOS and Android version of the Twitter application to add text description features, but the text description can not exceed 420 characters, hoping to let visually impaired people can also "see" the picture - by reading the text Description to achieve.