New Intel Device Promotes AI Algorithms, Computer Vision at Network Edge

Chipmaker Intel has unveiled a new version of its neural networks plug-in device aimed at helping developers reach further into the domain of artificial intelligence learning and edge computing.

The Neural Compute Stick 2 is effectively a USB stick containing the Movidius Myriad X Vision Processing Unit. The new unit is, in essence, a chip that performs eight times faster than previous stick versions, says Intel, and is designed to carry out accelerated computations related to computer vision and image recognition on network edge devices.

Intel said the stick could be usedOpens a new window to “prototype and deploy deep neural network applications smarter and more efficiently with a tiny, fanless, deep learning development kit designed to enable a new generation of intelligent devices.”

Developers insert the NCS 2 into a compatible USB 3.0 port on their computers and configure it with AI and computer vision know-how before slotting it into a smart device and testing its capabilities.

It’s not necessary to connectOpens a new window back to a central or cloud network because the stick can manage the inference of machine learning algorithms. It contains 16 powerful streaming hybrid architecture vector engine (SHAVE) processing cores and a deep neural network hardware accelerator for high-performance vision and AI inference applications, all running on a small amount of power.

As Jonathan Ballon, VPOpens a new window and general manager of the company’s Internet of Things group, says: “For the past four years or so, so much of AI has been happening in the cloud or data center, which is fine when you’ve got large volumes of data and you have unlimited compute resources, and power and cooling.”

Ballon was also keen to point out that outside the data center, the consumption of power, heat generated and running costs are extremely important.

Sticks have Pioneered Prototypes

The earlier iteration of the device, the Movidius Neural Compute Stick, was revealed a year ago and contained the Myriad 2 VPU. According to Intel, this earlier version revolutionized deep learning prototyping at the edge. It has been used in network edge devices such as smart cameras, including Google’s Clips and Tencent’s DeepGaze, and in DJI’s Phantom 4 drone.

The first device was also used in the Clean Water AI project, run by Intel software innovator Peter Ma, who came up with a method to identify bacteria in water using pattern recognition and machine learning. A digital microscope is linked to a laptop running the Ubuntu operating system and Intel’s compute stick, with contaminated water sites then being plotted on a map in real time.

Other projects that Intel highlights include cameras used in screening for skin and breast cancers, as well as spotting sharks from drones. Ballon confirms that the stick was practically designed for use in testing smart devices, industrial robots and drones.

Perusing the Deep Learning Edge Potential

Intel hopes the developer community will call on the NCS 2, compatible with the Ubuntu 16.04, CentOS 7.4 and Windows 10 operating systems, for further innovations using an open source software project called OpenVINO toolkit. The toolkit works with machine learning frameworks like Facebook’s Caffe2 and Google’s TensorFlow.

The first iteration of the device sold out quickly, says Steen Graham, general manager of Intel’s IoT channels and ecosystem. “We’ve really had very little insight about what the demand was,” he said. “A lot of people train models in the cloud; a lot of people do inference in the cloud – but deploying deep learning or AI at the edge, we didn’t know what the stage was in the deep learning community, what their interest was.”

According to a new report from market intelligence group IDC, more than 40% of organizational cloud deployments will include edge computing and 25% of endpoint devices and systems will execute AI algorithms by 2022.

According to the report: “Enterprises’ service delivery/execution foundation will be based on cloud infrastructure and platform services that support hybrid/multicloud deployment, increasingly distributed across public cloud, hosted private cloud, local/on-prem locations, and numerous “edge” locations (with proximity to IoT devices and data sources such as handheld terminals, cell phones, wearables, switches, drones, TVs, planes, surveillance cameras, self-driving vehicles, and smart buildings)”.

The NCS 2 is priced at $99, a slight step up from the last version at $79, and can operate in temperatures from zero to 40°C.