Movidius Partners with Dahua Technology Merging AI, Machine Vision, and Surveillance

Subscribe To Download This Insight

2Q 2017 | IN-4579

Machine vision and artificial intelligence (AI) startup Movidius entered the technology spotlight after being acquired by Intel last year (September 2016). This was a well-calculated move by Intel to capitalize on what could possibly be the next evolutionary link for innovative machine vision applications: programmable and ultra-low power vision chip designs. One of the company’s core beliefs, which fueled its R&D, is that machine vision technology can only reach the next level by embracing the interdependency between algorithm, as well as design and chip architecture for edge devices. This means that the bulk of pixel processing should occur right beside the edge device, allowing for improved optimization and programmability, as well as performance. Suffice it to say that it was this innovative chip design that also allowed the company to proceed to a successful partnership with Google.

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.

 

Movidius Joins Machine Vision Innovators

NEWS


Machine vision and artificial intelligence (AI) startup Movidius entered the technology spotlight after being acquired by Intel last year (September 2016). This was a well-calculated move by Intel to capitalize on what could possibly be the next evolutionary link for innovative machine vision applications: programmable and ultra-low power vision chip designs. One of the company’s core beliefs, which fueled its R&D, is that machine vision technology can only reach the next level by embracing the interdependency between algorithm, as well as design and chip architecture for edge devices. This means that the bulk of pixel processing should occur right beside the edge device, allowing for improved optimization and programmability, as well as performance. Suffice it to say that it was this innovative chip design that also allowed the company to proceed to a successful partnership with Google.

Partnering with Dahua Technology

IMPACT


Movidius states that some of its primary target applications include smart devices, wearables, robotics, drones, augmented reality (AR) headsets, and virtual reality (VR) headsets. Even with the lack of smart home and automotive applications, both still secure quite a large addressable market for the consumer side of the Internet of Things (IoT). The company, however, did not stop there and is currently attempting to break into the larger governmental and public safety market by targeting surveillance cameras.

In April (2017), Movidius paired up with Dahua Technology, the global leading name in surveillance, to provide its advanced AI and machine vision chip to surveillance cameras in the governmental public security market segment, with possible applications in mid- to high-end surveillance to the private sector, as well. The resulting partnership boasts a combination of:

  • Movidius’s Myriad 2 Vision Processing Unit (VPU), which manages to tackle the manageable area footprint problem while still featuring a high-power processor for edge devices;
  • Intel’s own RealSense computer vision technology empowering the VPU; and
  • Dahua Technology’s 2 MP and possibly 4 MP HD- and IP-connected surveillance cameras.

This technology mixture will allow for neural networks applications, processing and detection capabilities, a significant improvement in facial recognition, behavioral and emotional analysis, gender/age detection, and crowd-monitoring, among others. 

Smaller but Vital Addressable Market

COMMENTARY


In the recently published market data report Biometric Technologies and Applications (MD-BIOM-105) ABI Research estimates that total IP-connected surveillance cameras are expected to reach 81.1 million shipments by 2022, with HD cameras reaching 49.1 million by the same year, leaving the currently dominant Analog cameras behind with 12.08 million shipments.

Of course, this does not mean that the entire IP and HD surveillance camera segment is actually an addressable market for Movidius. Pricing and applicability will decide future partnerships. Moving processing to the edge and making use of artificial neural networks (ANNs) is definitely a promising endeavor for near future governmental surveillance, public safety, as well as global security initiatives from INTERPOL. However, these state-of-the-art capabilities are not on the top of the priority lists for surveillance systems in the private sector, which are still tackling the IP-connected penetration rates and larger video data storage for surveillance systems for both indoor and outdoor cameras—the only exception to this rule being casinos and (to a smaller extent) the larger retail market. Machine vision coupled with advanced recognition and low-energy requirements for edge devices is certainly a potent combination that we expect to see more of in the years to follow. Yet, this begs the question: could this increase in processing power also strengthen botnet “armies” as was demonstrated numerous times in the past with surveillance cameras from Asian vendors?  

Services

Companies Mentioned