New multimodal platform has “perceptive intelligence” for apps including smart displays and cameras, video soundbars and more, company says.
Synaptics has rolled out an AI-embedded platform at CES called the VideoSmart VS600 family of multimedia system on a chip (SoC) offerings with a CPU, NPU, and GPU.
The VS600 family is designed with what Synaptics calls “human perceptive intelligence for a new generation of smart displays, smart cameras, video soundbars, set-top-boxes, voice-enabled devices and emerging computer vision IoT products.”
SEE: Special report: From cloud to edge: The next IT transformation (free PDF) (TechRepublic)
The VideoSmart VS600 family uses Synaptics’ neural network acceleration and processing technology (SyNAP), which the company says executes deep-learning models in its devices.
The AI functionality provides features such as user identification and behavioral prediction through video, as well as vision and voice and a higher level of privacy, security, and real-time interaction for end users, the company maintains.
To enable edge computing in AI applications, the VS600 SoC offerings integrate a high-performance neural network processing unit (NPU) that supports trillions of operations per second, according to Synaptics, which makes human interface hardware and software.
An integrated MIPI-CSI camera serial interface along with an advanced image signal processing engine provides the camera input for edge-based computer vision inferencing, the company said. This is designed to minimize the need for external components, Synaptics said.
Edge-based voice processing is provided by Synaptics’ far-field voice and customizable wake-word technology.
The proliferation of smart home devices is making consumers are more vulnerable than ever to cyber threats and calls for a different approach, said Blake Kozak, senior principal analyst of smart home and security technology at IHS Markit.
“Combined with the ongoing transfer of personal data associated with cloud processes, and the actual personal benefits of using smart devices, consumers are struggling to be inspired by promises of manufacturers and service providers,” Kozak said in a statement. “This has led to the increased need for edge-based processing of smart home systems.”
Home security cameras and smart speakers that historically relied on the cloud will soon have significant functionality done right at the edge, Kozak added. This will bring consumers more security and peace of mind, he said.
“Although some cloud interfaces will be needed for home automation features, especially connections with external sources such as asking for a weather forecast, it is expected that advanced processes will move from a cloud infrastructure to being done locally in the home—where device-to-device communication, computer vision, and machine learning will improve the user experience,” Kozak said.
Also see
Source: TechRepublic
