Chip giant Qualcomm has launched its first sytem-on-chip devices for the Internet of Things (IoT) aimed at integrating into smart cameras.
The two 10nm SoCs include advanced camera processing software, machine learning and computer vision software development kits (SDKs), WiFi connectivity and security technologies. The leading edge process technology allows hardware that support AI on the chip as well as reducing the cost.
The Vision Intelligence Platform combines cutting-edge processing and AI with superior imaging, which is key because so many IoT applications, of today and tomorrow, use visual information. This goes beyond security cameras to include everything from ovens to robotic vacuum cleaners and drones — any device that relies heavily on information that comes from a camera.
The processing of that information could be done in the cloud, but that takes resources and time. Because the camera itself has the intelligence, it can decide how to respond based on what it knows instead of waiting for video data to be sent to the cloud and analyzed. This edge computing gives faster processing, local control, better security and privacy, and the use of less network bandwidth. Integrating this technology will also push the IoT ecosystem forward, as developers move away from the cloud and focus on the capabilities of the device.
Processing data at the edge requires significant horsepower, and the Vision Intelligence Platform has eight cores, from multiple 2.5GHz 64bit Kryo 300 customised ARM cores, Adreno 615 GPU, Hexagon 685 Vector Processor, Spectra 270 ISP camera core, as well as DSPs for sensors and audio. Support for up to 2x2 802.11ac Wi-Fi with MU-MIMO, Bluetooth 5.1, a 3D Audio Suite and other audio technolgies including aptX gives a wide range eof embedded capabilities.
The Vision Intelligence Platform also integrates a software AI engine called the Snapdragon Neural Processing Engine (NPE). The NPE provides analysis, optimisation and debugging capabilities that help developers and OEMs port trained networks into the platform. The AI Engine is compatible with Tensorflow, Caffe and Caffe2 frameworks, Open Neural Network Exchange interchange format, Android Neural Networks API, and Qualcomm's own Hexagon Neural Network library, so developers have the freedom to use their preferred framework.
The Qualcomm QCS603 and QCS605 SoCs are currently sampling but power consumption figures are not yet available. A QCS605-based VR/360 camera reference design from camera OEM Altek is available today. A QCS603 based industrial security camera reference designs will be available later in 2018.
The Qualcomm QCS603 and QCS605 SoCs are currently sampling but power consumption figures are not yet available. A QCS605-based VR/360 camera reference design from camera OEM Altek is available today. A QCS603 based industrial security camera reference designs will be available later in 2018.
No comments:
Post a Comment