San Mateo, Calif.-based Movidius, which specializes in embedded machine-vision technology, together with Hikvision announced that Movidius’ Myriad 2 Vision Processing Unit (VPU) technology will be powering a new lineup of smart cameras. Among other things, Myriad 2 will be utilized for running cutting-edge Deep Neural Networks in order to perform much higher accuracy video analytics locally.

The companies stated that Deep Neural Networks are rapidly replacing traditional computer vision approaches to video analytics, allowing security systems to automatically detect anomalies such as suspicious packages, drivers distracted by mobile devices, and intruders trying to access secure locations. Running Deep Neural Networks has historically required devices to depend on additional compute in the cloud, but due to the ultra-low power Myriad 2 VPU, these advanced algorithms can now be run at the edge, inside cameras themselves.

“Advances in artificial intelligence are revolutionizing the way we think about personal and public security,” said Movidius CEO Remi El-Ouazzane. “The ability to automatically process video in realtime to detect anomalies will have a large impact on the way cities’ infrastructure are being used. We’re delighted to partner with Hikvision to deploy smarter camera networks and contribute to creating safer communities, better transit hubs and more efficient business operations.”

Because of Deep Neural Networks and stereo 3D sensing, Hikvision has been able to achieve up to 99 percent accuracy in its advanced visual analytics applications, according to the company. Some of these applications include: car model classification, intruder detection, suspicious baggage alert, and seatbelt detection. The Myriad 2 platform allows these functions to now be processed instantaneously onboard the camera, rather than being sent to the cloud for processing.

“There are huge amounts of gains to be made when it comes to neural networks and intelligent camera systems,” said Hikvision CEO, Hu Yangzhong. “With the Myriad 2 VPU we’re able to make our analytics offerings much more accurate, flagging more events that require a response, while reducing false alarms. Embedded, native intelligence is a major step towards smart, safe and efficiently run cities. We will build a long-term partnership with Movidius and its VPU roadmap.”

Visit www.movidius.com for more information.

Potential Applications

Movidius outlined some potential sectors and applications in which the neural network Myriad 2 can run.

Retail

• Traffic analysis: building heat maps of stores according foot traffic, allowing companies to place retail items in optimal locations, and redesign and improve areas of low traffic.

• Theft prevention: detecting when an object is removed from an area (for example, demonstration products). Flagging of loitering/casing behaviors, which are distinct from browsing behaviors,

• Age/gender analysis: tailoring offering to demographic.

Security

• Intruder detection: defining areas of the video feed that will trigger an alert if a person is detected in a defined region; also known as “invisible fence.”

• Unattended/dropped-item detection: flag staff to items left suspiciously in public places (airports/crowded events).

• Automatic tracking: keep subject in view without manual operation of thumbsticks controlling pan/tilt/zoom.

Traffic

• Car counting (traffic flow optimization).

• Color detection.

• Model detection (road utilization by vehicle type).

• Pedestrian detection (traffic flow optimization).

• License plate identification.

• Car identification when license plate is not present (for example, dents, scratches, unique identifying features).

Residential Security

• Intruder alert: recognizes humans, rather than just a “dumb” motion sensor that will react to cats/dogs, etc.