26 October 2016

Uniview Unveils Myriad 2 Powered Surveillance & Security Solutions

October 25, 2016 China International Community Public Safety Expo, Beijing: Security and Surveillance experts Uniview have today unveiled a new Security Machine Vision (SMV) strategy combining chip, algorithm, network architecture, and product in a four-tier strategy to achieve a comprehensive breakthrough in intelligent security and surveillance solutions. Based on Movidius vision processing technology embedded directly into the cameras themselves, and a companion server-side intelligent video storage system, Uniview has developed a global chip strategy for deploying advanced video analytics. Uniview debuted this technology in three new products: a facial recognition camera, a deep learning-based intelligent traffic camera, and a deep learning accelerator dubbed “Smart Stick”.

Security Machine Vision (SMV) adds increasing momentum to the adoption of the "visual intelligence of things" concept. Over the past year, the field of artificial intelligence has been moving at a rapid pace, and Uniview has been hard at work applying over a decade of experience into these new products. By adopting a SMV-focused strategy starting from the product core with specialized embedded chips and algorithms, Uniview is able to rapidly deploy new levels of intelligence in its security and surveillance solutions.

Architecture at the core: Smart Chain Computing

In 2006, Uniview launched IP video monitoring with IP network full-exchange architecture to solve the problem of urban network monitoring. In 2016, the introduction of intelligent chain computing architecture, with a "high-speed all-media bus", opens up the intelligent monitoring to achieve "vertical and horizontal" resource linking and collaboration between cameras. This configuration significantly improves recognition accuracy, resource utilization and system performance. This structural network innovation is designed to meet the challenges of large-scale intelligent monitoring and the real-life demands large-scale deployments.

Chip + Algorithm: Bringing Intelligence to the Edge

The Movidius Myriad 2 Vision Processing Unit (VPU) was selected by Uniview as the front-end processor for the new SMV strategy. As the industry-leading processor when it comes to deep neural networks, the Myriad 2 chip will power a family of high-end cameras and smart stick products. Myriad 2 can run deep learning algorithms at extremely low power, enabling front-end face recognition, behavior analysis, vehicle identification and other advanced video analysis applications.

Thanks to the recent progress made by professional research institutes and top university laboratories in the field of deep learning, Uniview has been able to deploy video analytics with up to 25 percent greater accuracy than traditional back-end based approaches.

Uniview is demonstrating it's newest technology at the China Public Safety Expo in Beijing October 25th through 28th.