Monday, November 7, 2016

Qualcomm launches edge-analytics IoT camera

From the research site Rethink Wireless:

The most interesting new product was the camera design, which has dropped the branding used in the briefing materials, for some reason. The camera system aims to bring network-edge analytics to video processing, as well as acting as a sensor hub for IoT applications.
In terms of spec, the camera is capable of HEVC encoding at 4K resolutions at 30fps, powered by Qualcomm’s Snapdragon 625 SoC. With LTE, WiFi and Bluetooth, as well as I/O pin-outs on the back for those additional IoT sensors, Qualcomm is providing a platform reference design for developers to take and turn into products – driving Qualcomm sales and orders.
The 8-core CPU on the Snapdragon 625 is going to be providing the processing power needed for running video analytics on the device, with will allow developers to reduce the network bandwidth used in backhauling video for analysis in a central cloud deployment.

This article focuses on single definition of Edge Analytics.  The type of analysis done by these edge-analytics cameras is  Much like Intel, I think the emergence of Smart Cameras will be a huge boon to performing analytics against image data, as it alleviates bandwidth concerns (and possibly improves security as they will probably be more easily updated).

There is a huge opportunity for edge analytics when using cameras, as you still need to manage the cameras, correlate events from multiple cameras and use that as input to a cloud service.

Our goal with our ViznTrac Edge concept is to provide a simple way of detecting and managing inputs from IP cameras on a local network and have that send the relevant data to our ViznTrac service.  I find articles such as the one above to be very optimistic about the opportunities available in image analytics.