Ambarella has unveiled the CV5 AI Vision Processor to handle 8K image processing for car, consumer, crone, and robot cameras.
The new 5-nanometer camera system on chip (SoC) is the latest in the company’s CVflow family, and it consumes under 2 watts of power. The chip is about 4 times more powerful than the previous-generation chip.
Chris Day, VP of marketing and business development at Ambarella, said in an interview that the chip can record 8K video or four 4K video streams. Samples will be available in the second quarter, and products using the chips may be out in 2022.
The chip will enable the development of intelligent automotive camera systems, consumer cameras (drone, action, and 360°), and robotic cameras. The company made the announcement as part of the CES 2021 digital trade show this week.
The chip combines Ambarella’s CVflow AI engine with dual Arm A76 CPUs to provide the performance necessary for a wide range of AI-based algorithms. Its advanced image signal processor (ISP) can simultaneously process 8K images at 30 frames per second for both human viewing and machine processing.
“In sports cameras, power consumption is critical because of the compact form factor, and they run out of batteries,” Ambarella senior marketing director Senya Pertsel said in an interview with VentureBeat. “In particular, we emphasize HDR improvements for challenging lighting scenarios, as well as on-chip image stabilization for sports cameras.”
The chip is made in a 5-nanometer process technology, where the width between circuits is five billionths of a meter. That technology enables Ambarella to achieve the combination of high performance and low power consumption.
“Five nanometers is very much the leading edge of technology,” Day said.
In November, Ambarella unveiled the CV28M low-end camera chip.
“CV28M was the baby of the family for low-cost sensing cameras,” Day said. “And CV5 is the other extreme as our highest video performance chip. We’ve been cranking out different chips with different AI capabilities.”
In automotive video telematics applications, CV5 provides the performance necessary to encode multiple video streams from front safety features, driver monitoring, cabin monitoring, and side-view cameras.
“The [8K] performance is already useful. First, there is a balance between resolution and frames per second with the maximum dictated by this performance so even if the application does not make use of 8K, the additional resource can be applied to increase the FPS, which is commonly useful,” Yole Développement semiconductor analyst Tom Hackenberg said in an email to VentureBeat. “Second, while we might view 8K as overkill for visual displays in embedded applications, the additional data may allow for greater accuracy when inferencing the content of the video.”
The CVflow AI engine can simultaneously run advanced driver assistance system (ADAS) algorithms, such as whether the driver is drowsy or is veering out of a lane.
The combination of high-resolution video capture with advanced AI processing enables ADAS cameras to recognize images over long distances and with high accuracy.
“In the drone market, CV5 brings next-generation videography and cinematography by supporting 8K video with excellent image quality and compression efficiency,” Pertsel said. “It will enable things like object detection, obstacle avoidance, sensor fusion, and path planning for autonomous drones.”
For drone-based aerial videography and cinematography, CV5 can simultaneously perform flight control and navigation functions while recording up to 8K resolution video recording at 60 frames per second.
“We think multi-camera systems will be one of the key applications, like having lots of cameras on a semi-truck,” Pertsel said.
The CV5 SoC shares common SDK and computer vision (CV) tools with Ambarella’s other CVflow system-on-chip (SoC) families, simplifying development of cameras with multiple price and performance options. It has common tools for developers.
I asked if it was overkill to have an 8K video of someone perhaps falling asleep in a car. Ambarella marketing director Udit Budhia said in an interview that semi-truck monitoring might need such image processing power because it would have so many cameras keeping track of different parts of the truck, including the part where a camera monitors whether a driver is drowsy. All of that could add up to requiring 8K processing.
The Santa Clara, California-based company, which went public in 2012, started out making low-power chips for video cameras. But it parlayed that capability into computer vision expertise and in 2018 launched its CVflow architecture to create low-power artificial intelligence chips. Now the company has 800 employees and is competing with the likes of Intel and Nvidia, though with a focus on low-power applications. Ambarella generated $220 million in revenue in 2019.
VentureBeat’s mission is to be a digital townsquare for technical decision makers to gain knowledge about transformative technology and transact.
Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you,
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform
- networking features, and more.
Author: Dean Takahashi