xi's moments
Home | Newsmakers

AI apps in smart glasses to help visually impaired

By Barry He | China Daily Global | Updated: 2022-05-05 09:54

Being able to see is an ability millions of people around the world lose each year. Without our primary sense, maintaining independence and quality of life can be a struggle, but technology is now producing several exciting ways in which to aid the visually impaired.

AI programs in smart glasses can provide context to their environment, automatically reading out audio renditions of text, or even describing what is seen by the user in real time. Other devices manipulate captured images in real time and clarify them, providing images for those with some residual degree of sight.

Traditionally, technology employed to help the visually impaired has had a heavy emphasis on office-related IT tasks, for example, reading or using a computer. While enabling these individuals to better engage in the economy and offer their talent in the workplace, such solutions do little to offer context for real-world scenarios.

This is now changing, as AI programs allow smart glasses to describe trees or plants in a park, and even recognize pre-programmed faces of friends and family. The ability to recognize objects, colors and faces with nuanced navigation between each real-world scenario means that technology is empowering the differently abled like never before.

Just like smartphones, smart glasses of the future will host app ecosystems, allowing specialist external services to broaden the utility of the device. Apps such as the Cash Reader app already allow certain smart glasses models to recognise more than 100 different types of currency, making travel easier for the visually impaired. Words or even audio phrases may also be translated into other languages in real time, meaning that the technology may also have crossover mainstream appeal to those who are not visually impaired.

The advent of 5G also means that video connections showing a live feed of a wearer's point of view can be directly streamed in real time to a trusted friend or relative. Fast internet speeds with low latency mean that the user can be guided through a scenario or have a situation explained to them near-instantaneously, with specific context relating to the user they know individually.

A common misconception is that the blind see nothing, or blackness. However, this is far from the case. The vast majority of blind people do have some remaining degree of vision, and smart glasses that can intelligently clarify images for the user are becoming increasingly popular. As the technology becomes more sophisticated, programs can identify visual cues to increase the contrast or brightness of images fed to the user. Irrelevant objects can be removed to prevent distractions, with hazards highlighted in black and white. Doorways and corridors can also be enhanced, while human faces are automatically recognized by AI, translating them into a crisper, focused image. It is only in the last few years that the technology has reached a level where devices can understand the nuance of fast-paced and detailed environments to make such smart glasses feasible for users.

Object recognition is a key product of deep learning and machine learning algorithms that makes such innovation possible. Humans can achieve this with ease; however, training a computer to spot objects and scenes and to understand their context is something that has until only recently been possible artificially.

Systems are improving rapidly, and as they do, the visually impaired will find themselves becoming independent with a quality of life like never before.

Barry He is a London-based columnist for China Daily.

Global Edition
Copyright 1995 - . All rights reserved. The content (including but not limited to text, photo, multimedia information, etc) published in this site belongs to China Daily Information Co (CDIC). Without written authorization from CDIC, such content shall not be republished or used in any form. Note: Browsers with 1024*768 or higher resolution are suggested for this site.
License for publishing multimedia online 0108263

Registration Number: 130349