New OmniVision Sensor Can Recognize Faces in the Dark

“OmniVision’s solution circumvents the need for such an elaborate approach, and the company says it’s highly energy efficient to boot.”

A new image sensor could help to extend the reach of facial recognition technology to nighttime applications.

The solution, dubbed “OS02C10”, comes by way of OmniVision Technologies. It’s a 2.9-micron, 2MP image sensor featuring what OmniVision calls ‘ultra-low-light’, or ‘ULL’, technology. The latter, combined with OmniVision’s Nyxel near-infrared technology, allows for strong imaging across a wide range of lighting conditions – all of them, in fact, according to the company.

It can even detect light in conditions too dark for the human eye, which “means that indoor security cameras will not disturb sleeping occupants, and can be easily concealed from intruders,” according to a statement announcing the solution.

That, of course, leads to compelling surveillance applications. As OmniVision business development director Brian Fang explains, “Surveillance cameras with AI functionality need the highest possible resolution in all lighting conditions for accurate facial recognition.” Indeed, this is a technical challenge that has so far required some real ingenuity, with Army researchers having announced a technique earlier this year for converting thermal images to conventional pictures upon which facial recognition can be applied.

OmniVision’s solution circumvents the need for such an elaborate approach, and the company says it’s highly energy efficient to boot. That could be appreciated not only in the battlefield, but across a wide range of security applications.

–

October 16, 2018 – by Alex Perala