January 5, 2022
Ambarella, which develops artificial intelligence vision silicon, has announced its new Artificial Intelligence Image Signal Processor (AISP). The company is demonstrating this technology at this week’s CES 2022 event in Las Vegas, running on its CV2 edge AI perception systems-on-chip (SoCs).
Developed from its CVflow AI engine, the new AI-based ISP architecture uses neural networks to augment the image processing done by the hardware ISP integrated into its SoCs, Ambarella said. This approach enables color imaging with low light at very low lux levels and minimal noise, which is a 10x to 100x improvement over traditional ISPs, and new levels of high dynamic range (HDR) processing with more natural color reproduction and higher dynamic range.
The company said the new AISP will become available across its entire CVflow SoC portfolio at resolutions up to 4K. Applications for the new technology include markets where low-light and HDR processing are critical, including security and automotive cameras, where full-color night vision is a key operating point, and where better low-light and HDR enable reductions in external illumination, the use of lower cost sensors, and cleaner video for human viewing and analytics processing.
An example of this is a security camera trying to identify a license plate at night, the company said. “With noisy video, the text on the license plate would be unreadable, even to the best AI algorithm,” the company said. “And with black and white video, all color information on the car would be lost. By using Ambarella’s AISP technology and restoring colors at the source before AI processing, the license plate can be clearly read and the color of the car identified.”
“Being able to see clearly in low-light or high-contrast conditions is key to robust camera systems,” said Les Kohn, Ambarella’s co-founder and CTO. “Traditional camera systems have had to live with noisy or dark black and white video in low-light conditions, and dark shadows or blown-out highlights in high contrast conditions. Both of these cases result in the loss of details that are detrimental to both human viewing and AI applications. With our new AISP technology, we increase the useful range of camera systems while reducing the total system cost to build high quality cameras.”
Ambarella said the AISP processing happens in parallel on the silicon with other higher-level AI applications running on the CVflow AI engine. This lets customers build their next generation of cameras with both higher visual quality and more accurate AI algorithms, while staying with the low power budgets required by small embedded systems.
The company said it expects the AISP to be made available during 2022, with different prices and performance options.
Ambarella also announced the CV3 AI domain controller family, a series of SoCs aimed at the automotive industry for AI processing performance at up to 500 eTOPS, a 42x increase over its previous automotive family of SoCs.
For more details on the company and its technology, visit the Ambarella website here.