Ubicept's technology can improve machine vision of existing cameras as well. Ubicept
Ubicept, a US-based startup is deploying technology used in iPhone-LiDAR to improve machine vision even in variable lighting conditions. Showcased today at the ongoing CES 2025, the technology can help overcome hurdles faced by sectors such as augmented/ virtual reality, autonomous vehicles, and robotics.
Cameras and image sensors have come a long way from their preliminary forms, developed in the late 1900s. From capturing only 0.01-megapixel images in black and white, sensors can now capture 400-megapixel images with high color reproduction.
Alongside the improvement in image capturing, the devices have also shrunk and now easily fit into devices such as smartphones. Yet, there are multiple limitations to using them in challenging light conditions. For instance, capturing images of fast movements at night results in images that are too blurry or noisy for even advanced technologies like artificial intelligence (AI) to correct.
A team of computer vision experts from institutes such as MIT, the University of Wisconsin-Madison, and others who have previously worked at Google and Facebook have come together at Ubicept to solve these issues with technology that uses photon-level image data processing for improving machine perception clarity.
iPhone-LiDAR tech
“The next big AI wave will be enabled by computer vision-powered applications in the real world; however, today’s cameras were designed for humans, and using standard image data for computer vision systems won’t get us there,” said Tristan Swedish, cofounder and CTO at Ubicept in a press release.
“Ubicept’s technology bridges that gap, enabling computer vision systems to achieve ideal perception.”
Ubicept’s approach to taking camera vision to the next level involves using single-photon avalanche diode (SPAD) sensors, the technology used in iPhone’s LiDAR technology, to create unified imaging solutions that remove the need for different specialized cameras.
The technology allows simultaneous handling of bright and dark areas in the same environments while capturing high-speed motion without any light streaks, generating crystal clear images even in extremely low-light environments.
“By processing individual photons, we’re enabling machines to see with astounding clarity across all lighting conditions simultaneously, including pitch darkness, bright sunlight, fast motion, and 3D sensing,” added Sebastian Bauer, cofounder and CEO of Ubicept in the press release.
Sample of image enhancement achieved using Ubicept’s technology. Image credit: Ubicept
Enabling larger adoption
To facilitate larger adoption of this technology, Ubicept is rolling out the Flexible Light Acquisition and Representation Engine (FLARE) Camera Development Kit that allows camera companies, sensor makers, and computer vision engineers to integrate its technology into autonomous vehicles, robotics, AR/VR, industrial automation, and surveillance applications.
While this kit comes with a 1MP camera with a full-color SPAD sensor, Ubicept is also open to integrating its technology into existing cameras and CMOS sensors to advance any camera into a next-generation advanced imaging system.
“SPAD cameras are revolutionizing low-light imaging with photon counting for unmatched performance, limited to niche applications,” explained Florian Domengie, Principal Analyst, Imaging from Yole Group, in a press release sent to Interesting Engineering.
“Yet, adoption in broader markets such as industrial, automotive and consumer faces challenges like pixel size limitation and high data processing and computation load. Companies like Ubicept are tackling these issues with advanced algorithms, enabling low latency, high frame rate and dynamic range for future wider applications.”
Ubicept’s breakthrough in machine perception was showcased today at CES 2025.
