According to Microsoft’s official blog, the company has developed a new image-captioning AI algorithm. The company hopes to use artificial intelligence to achieve barrier-free products.
In some limited tests, the accuracy of this new image algorithm has surpassed that of humans, and it will obviously help Microsoft to achieve the accessibility of its own products.
Microsoft launched the Seeing AI application in 2017, which uses artificial intelligence algorithms to identify the content taken by the user and describe it in language.
The new image algorithm will also be provided to the Seeing AI application to improve its item recognition rate and help visually impaired people understand the world through the camera.
The Microsoft Office software suite currently provides some accessibility measures. For example, the accompanying screen reading software can read the content line by line to help visually impaired people identify.
The application of the new image subtitle algorithm in office software is to create alternative text for pictures, which can be recognized by screen reader software and converted into voice content.
The so-called alternative text mainly refers to the ALT attribute of the picture, which was originally used to describe the content of the picture, but the text content depends on the user to edit it manually.
The new algorithm can automatically recognize the content of the picture and create alternative text. The alternative text can be handed over to the screen reader software to become voice so the visually impaired can understand the content.
It is reported that the above new functions will gradually arrive in Word, PowerPoint, Outlook, and other components, and other office software components are expected to support this function.