Real-time Video Analytics: the killer app for edge computing
- Ganesh Ananthanarayanan ,
- Victor Bahl ,
- Peter Bodík ,
- Krishna Chintalapudi ,
- Matthai Philipose ,
- Lenin Ravindranath Sivalingam ,
- Sudipta Sinha
IEEE Computer |
According to a 2015 report by IHS on the installed base for video surveillance equipment, there is a camera installed for every 29 people on the planet, with mature markets like the US having a camera for every eight of its citizens. The report expects the number of cameras to grow by 20% year-over-year for the next five years. Video analytics from these cameras are used for traffic control, retail store monitoring, surveillance and security, as well as consumer applications including digital assistants for real-time decisions.
Our position is that a geo-distributed architecture of public clouds, private clusters and edges extending all the way down to compute at the cameras is the only viable approach that can meet the strict requirements of real-time, large-scale video analytics.
Besides low latency and efficient bandwidth usage, another major consideration for continuous video analytics is the high compute cost of video processing. Because of the high data volumes, compute demands, and latency requirements, we believe that cameras represent the most challenging of “things” in Internet-of-Things, and large-scale video analytics may well represent the killer application for edge computing.