IntelliVision introduces video analytics for Ambarella CV22 SoC I
09 May 2019 14:06 GMT

IntelliVision Technologies has introduced its video analytics for the Ambarella CV22 IP Camera SoC, based on Ambarella’s new CVflow architecture which includes hardware support for deep neural networks.

IntelliVision demonstrated its human and vehicle detection, license plate recognition, and face detection and recognition on the CV22 last month at ISC West 2019 in Las Vegas. IntelliVision video analytics will be available on the CV22 within 30 days.

“IntelliVision has been a close partner of Ambarella’s for many years,” said Chris Day, Ambarella’s VP of Marketing and Business Development, “and has optimized their video analytics including LPR, facial recognition and ADAS for our SoCs. We’re glad to see them take advantage of our new neural network support in the CV product line.”

The CV22 combines image processing, 4Kp30+ video encoding and CVflow computer vision processing in a single, low power design and is designed to run AI and computer vision applications. It provides specific support for deep learning and neural networks resulting in performance improvements of 10-100X over previous CPU-based embedded platforms.

“We’ve been blown away by the run time performance improvements in our CNN models provided by the new CVflow architecture,” said Krishna Khadloya, VP Development at IntelliVision. “It meshes perfectly with our neural network AI-based video analytics and will open up the edge to even more intelligent video and audio processing.”

The IntelliVision family of video analytics solutions includes object detection and classification (humans, vehicles, pets, airplanes), people/vehicle counting, demographics, intrusion/perimeter watch, object left/removed, license plate recognition (ALPR/ANPR), audio/sound recognition and facial recognition.