Smart Glass Could Offer a New Route to Machine Vision

Facebook X LinkedIn Email
A new type of smart glass, developed by a team at the University of Wisconsin-Madison, leverages optical reflection to recognize images without requiring sensors, circuits, internet connection, or external power sources. Everything needed for image recognition is condensed into single pieces of glass. The technology could be used someday to embed facial recognition in smartphones. The glass distinguishes among different images by distorting incoming lightwaves.

Researchers Zongfu Yu (left), Ang Chen (center), and Efram Khoram (right) developed the concept for a smart glass that recognizes images without any external power or circuits. Courtesy of Sam Million-Weaver.
Researchers Zongfu Yu (left), Ang Chen (center), and Efram Khoram (right) developed the concept for a smart glass that recognizes images without external power or circuits. Courtesy of Sam Million-Weaver.

The glass works as a nanophotonic media that can perform nonlinear mode mapping in a way that is comparable to artificial neural computing. Complex information is encoded in the wavefront of an input light. When the lightwave enters the glass, the glass transforms the wavefront to perform computing tasks such as image recognition. 

Computation is performed by a host material with numerous inclusions that appear as bubbles and impurities embedded in the glass. These inclusions scatter light in both forward and backward directions to differentiate images. The scattering mixes the input light in a manner that can be compared to linear matrix multiplication in a digital artificial neural network (ANN). The locations and shapes of the inclusions can be compared to weight parameters in digital ANNs. At the output, the optical energy is concentrated in well-defined locations, which, for example, can be interpreted as the identity of the object in the image. The researchers believe that their smart glass could open up a new, analog approach to machine learning.

The computation performed by the smart glass is intrinsic to the material, so one piece of glass could be used indefinitely. “We could potentially use the glass as a biometric lock, tuned to recognize only one person’s face,” professor Zongfu Yu said. “Once built, it would last forever.”

For their proof of concept, the researchers created glass pieces that could identify handwritten numbers. Light emanating from an image of a number entered at one end of the glass; then it was focused on one of nine spots on the other side of the glass. Each spot corresponded to a different number. The smart glass was dynamic enough to detect in real time when a handwritten “3” was altered to become an “8.” Training the glass to recognize numbers was similar to machine learning, the researchers said, except that training was being given to an analog material, instead of a digital code.

Smart glass mimics artificial intelligence by bending light to recognize different numbers. Courtesy of Zongfu Yu, University of Wisconsin-Madison.
Smart glass mimics artificial intelligence by bending light to recognize different numbers. Courtesy of Zongfu Yu.

“We’re accustomed to digital computing, but this has broadened our view,” Yu said. “The wave dynamics of light propagation provide a new way to perform analog artificial neural computing.”

In the future, the researchers plan to determine if their approach works for complex tasks such as facial recognition. Unlike human vision, which can discern a large number different objects at once, the smart glass could excel in specific applications — for example, one piece for number recognition, a different piece for identifying letters, and another for faces.

“The true power of this technology lies in its ability to handle much more complex classification tasks instantly without any energy consumption,” Ming Yuan, Columbia University professor and collaborator on the research, said. “These tasks are the key to create artificial intelligence — to teach driverless cars to recognize a traffic signal, to enable voice control in consumer devices, among numerous other examples.”

Although the upfront training process could be time-consuming and computationally demanding, the glass itself is easy and inexpensive to fabricate. 

“We’re always thinking about how we provide vision for machines in the future and imagining application-specific, mission-driven technologies,” Yu said. “This changes almost everything about how we design machine vision.”

The research was published in Photonics Research, a publication of OSA, The Optical Society ( 

Simple smart glass could offer a different route to artificial vision. Courtesy of University of Wisconsin-Madison.

Published: July 2019
embedded vision
Embedded vision refers to the integration of computer vision technologies into various embedded systems, devices, or machines. Computer vision involves teaching machines to interpret and understand visual information from the world, much like human vision. Embedded vision takes this concept and applies it to systems where the processing occurs locally within the device, as opposed to relying on external servers or cloud-based services. Key components of embedded vision systems include: ...
machine vision
Machine vision, also known as computer vision or computer sight, refers to the technology that enables machines, typically computers, to interpret and understand visual information from the world, much like the human visual system. It involves the development and application of algorithms and systems that allow machines to acquire, process, analyze, and make decisions based on visual data. Key aspects of machine vision include: Image acquisition: Machine vision systems use various...
machine learning
Machine learning (ML) is a subset of artificial intelligence (AI) that focuses on the development of algorithms and statistical models that enable computers to improve their performance on a specific task through experience or training. Instead of being explicitly programmed to perform a task, a machine learning system learns from data and examples. The primary goal of machine learning is to develop models that can generalize patterns from data and make predictions or decisions without being...
An SI prefix meaning one billionth (10-9). Nano can also be used to indicate the study of atoms, molecules and other structures and particles on the nanometer scale. Nano-optics (also referred to as nanophotonics), for example, is the study of how light and light-matter interactions behave on the nanometer scale. See nanophotonics.
Research & TechnologyeducationAmericasAsia-PacificUniversity of Wisconsin-MadisonImagingfacial recognitionsmart glassLight SourcesMaterialsOpticsEmbedded Visionmachine visionmachine learningneural networksanalog artificial neural networksConsumernanoThe News Wire

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.