If your passenger falls asleep in a car, it’s not a big deal. But if the driver falls asleep, lives are at risk. This is why AI chipmaker Ambarella and eye-tracking camera maker Smart Eye are teaming up to create technologies that sense what’s happening inside the cabin of a car.

Smart Eye said at CES 2019, the big tech trade show in Las Vegas this week, that it will use Ambarella‘s new CV22AQ CVflow computer vision processor in its in-cabin monitoring systems.

Santa Clara, California-based Ambarella is using its position in camera sensors and processors to gain a foothold in computer vision and artificial intelligence chips for driver-assistance cameras, electronic mirrors, in-car cameras, and parking assistance technology. The company is debuting the new CV22AQ and CV25 chips at CES.

Above: Ambarella provides a variety of chips for smart home applications.

Image Credit: Ambarella

Chris Day, chief marketing officer at Ambarella, said in an interview with VentureBeat that the company is moving beyond the consumer imaging market and focusing on security, surveillance cameras, and automotive cameras.

“We’ve gotten very good feedback on our CVflow computer vision architecture,” Day said. “We are new in the automotive market, which moves at a slower speed than security and surveillance. We are in evaluation at a lot of different companies.”

The car cameras will be able to track eye, mouth, and head position and movements to see if a driver is getting drowsy. (I imagine what follows is some kind of electric shock that will jolt the driver into consciousness). Smart Eye’s Driver Monitoring System tracks the driver’s actions and intentions, using the Ambarella chip, which delivers high-performance AI at just 2.5 watts of power consumption.

“We are very pleased to be working with Ambarella to enable advanced AI in the next generation of compact driver and in‐cabin monitoring camera designs,” said Martin Krantz, CEO of Smart Eye, in a statement. “The pairing of Ambarella’s CVflow high‐performance, low power-consumption computer vision processing with Smart Eye’s growing array of high‐accuracy and AI‐based driver monitoring algorithms offers a highly effective, scalable solution for Smart Eye’s OEM and tier‐1 customers. With the Ambarella CV22AQ, Smart Eye is able to provide high‐resolution, high‐precision head pose, gaze, eyelid, and mouth tracking in 60Hz, paired with concurrent execution of our growing portfolio of AI‐based interior sensing algorithms.”

Above: Ambarella’s latest computer vision chips.

Image Credit: Ambarella

“We are seeing significantly increased demand for both driver and in‐cabin monitoring cameras,” said Ambarella CEO Fermi Wang in a statement. “Powered by CV22AQ, this joint platform will allow system designers to fully optimize Smart Eye’s innovative tracking technology in high-performance, low-power system designs.”

The Ambarella CV22AQ offers support for both global shutter and rolling shutter CMOS sensors, both of which are required for in‐cabin applications. The processor’s Image Signal Pipeline (ISP), with support for RGB‐IR color filter arrays, enables accurate detection and monitoring, even in low‐light cabins. Its High Dynamic Range (HDR) processing extracts maximum image detail in high‐contrast scenes, further enhancing the computer vision capabilities of the chip and performance potential of Smart Eye algorithms, which figure out how alert someone is.

Smart Eye is based in Gothenburg, Sweden and has more than 700 clients around the world, including the U.S. Air Force, NASA, BMW, Lockheed Martin, Audi, Boeing, Volvo, and GM.

In separate news, Ambarella introduced its CV25 SoC chip with CVflow computer vision for a new generation of intelligent cameras. It is targeted at home security, professional surveillance, and aftermarket automotive cameras and uses deep neural network processing for applications like smart dash cameras, driver monitoring systems, and electronic mirrors.

In smart video doorbells, CV25 can automatically recognize familiar faces approaching the front door, flag unknown persons, and alert the homeowner when a package is delivered. In driver monitoring systems, it can detect a driver’s drowsiness or level of distraction by monitoring their eyes and facial expressions.