Imagia has announced a significant advancement in optical metasurface technology with the development of Processing Optics, which facilitates optically-accelerated feature detection. This new technology performs image processing tasks without requiring power or code. It uses an array of microscopic optical filters to identify intricate features such as human hands or faces, replacing the conventional on-chip algorithmic image processing methods. This innovation is particularly beneficial for power-limited and latency-sensitive applications, such as edge computing on wearable devices, as it allows for rapid and efficient feature extraction. This advancement can potentially expand the design possibilities for AI applications at the edge.
Normally, image detection and recognition can be a GPU or CPU intensive exercise. This means that whatever device is handling the recognition of faces, people, vehicles, or other items needs to have some processing power dedicated to detecting the items. Imagia’s new tech goes a layer deeper than this, letting the actual physical optical lenses determine if an object is found or not.
The foundation of this technology is based on five years of research in metamaterial optics conducted by Imagia’s team of scientists and engineers. Initially focused on developing metalens technology for optical components in devices like AR/VR headsets, Imagia has now extended its work to perform mathematical convolutions directly in optical elements. The system functions by applying mathematical convolutions through a series of optical filters. As light passes through a metalens, it is manipulated by billions of nanoscale components, which encode a pattern recognition algorithm directly into the optical signal. This process allows for a remarkable reduction in detection latency, achieving a 500-fold decrease compared to traditional methods, while using significantly less power.
Imagia has demonstrated the capabilities of this technology with a hand and gesture detector that operates using only eight pixels of information, responding in just 80 microseconds. In comparison, typical optical and processing techniques require 30-40 milliseconds to process millions of pixels. By processing images directly through optics, Imagia can achieve a comparable framerate to standard systems while consuming less than 1% of the power. This approach has significant implications for applications such as artificial intelligence and active feature detection in devices like laptops and AR/VR headsets, as it could lead to a 20% or more increase in battery life.
The company has launched an early access program to allow partners to investigate solutions and applications on the Processing Optics platform. This initiative follows successful demonstrations of Processing Optics to multinational device manufacturers in various markets, including semiconductors and consumer electronics. Imagia plans to release its first product, a detector module for mobile devices utilizing Processing Optics, in 2025.
Imagia is a deep tech startup focused on applying metamaterial principles to develop high-performance optical lenses and filters compatible with visible, incoherent light. Imagia’s optics are manufactured using standard CMOS processes, enabling ultra-fast and ultra-low-power feature detection for next-generation AI devices. The company’s meta-optics open up new possibilities in designing optical devices and imaging systems that were previously unattainable.
For more information, visit Imagia’s website here.
