Novel Chip Can Process Two Billion Images Per Second
Artificial intelligence can be used to improve a wide range of systems, but this also entails additional hardware requirements. To replicate the ability of biological neural networks to recognize or classify new data points such as images, robust hardware with enhanced speed and capabilities is needed.
Scientists from the University of Pennsylvania School of Engineering and Applied Science (Penn Engineering) have developed a chip that, despite its minuscule size of just 9.3 square millimeters, can detect and classify an image in less than a nanosecond, without the need for a separate processor or memory unit.
Speeding up the ability of any computer to process images is very important for many applications, such as face recognition algorithms, automatically detecting text in photos, or even enabling self-driving vehicles to recognize obstacles faster and better.
The new chip made by the Penn engineers not only classifies and recognizes images significantly faster than conventional chips, but is also scalable. The new level of performance is achieved by using an optical deep neural network that directly processes the light received from the object of interest.
“Our chip processes information through what we call ‘computation-by-propagation,’ meaning that unlike clock-based systems, computations occur as light propagates through the chip,” said Firooz Aflatouni, a member of the research team. “We are also skipping the step of converting optical signals to electrical signals because our chip can read and process optical signals directly, and both of these changes make our chip a significantly faster technology.”
Furthermore, as Aflatouni notes, eliminating the memory unit that stores images also increases data privacy. “With chips that read image data directly, there is no need for photo storage and thus, a data leak does not occur.”
Currently, the team is exploring the scalability of the chip and also further developing its three-dimensional object classification capabilities.
Did you enjoy this article?
You can find more on our Supertrends Pro app. Sign up for a 30-day free trial and find all the innovations and scientific discoveries relevant to you and your organization in one place.