Video Content Analysis: The Last Mile is Key

Intelligence on Edge
Therefore, many solution providers are moving the processing elsewhere and are looking to embed intelligence in their cameras, Gorski said. “There are many benefits in incorporating analytics onto the edge. This approach dramatically reduces bandwidth and storage requirements by eliminating the need to send irrelevant video data across the network to a centralized server. Decentralization enhances efficiency.”

For some applications, intelligence on the edge is a must, Otterspeer added. “Raw data is processed to make sure everything is analyzed. It is also much more scalable. When you add a camera, you are also adding processing horsepower, as opposed to constant upgrades in a centralized environment.” Power usage of a smart network camera is also significantly less than that of a centralized server.

When there is no need for a centralized server, licensing expenses in the server's operating system can be saved, Otterspeer said. “In the long run, this reduces the TCO for the end user.”

Process ing No Longer A Bottleneck
Looking beyond surveillance, cellular and mobilede vices are increasingly powerful and affordable. “Consumers

today are buying millions and millions of mobile devices; as a result, chip costs are coming down, creating a great opportunity for VCA,” Otterspeer said. “Having a dedicated chip for VCA allows the camera to be capable of ensuring optimal image quality, while delivering the full potential of VCA. There need not be a compromise.”

More powerful and affordable processors, in turn, drive more complex VCA features into cameras, said Patrick Lim, Director of Sales and Marketing for Ademco Far East. “Hopefully, cameras will become increasingly smarter. They could adjust intelligently for diverse lighting and weather conditions as well as nonenvironmental patterns, such as a diverse range of human activities, to assist in preemptive intelligence.”

●Ed Troha,MD of Global Marketing, ObjectVideo

“ Typically, ' lighter ' VCA implementations are seen on the edge to provide more basic functions, whereas more processingdemanding algorithms and applications still get deployed centrally,” Bohn added.

If the software is optimized, the VCA features should have very little effect on the performance of the processor, added Ed Troha, MD of Global Marketing at ObjectVideo.

“If the software is inefficient, it will take up excessive resources no matter how powerful the processor is.” Despite all the processor developments, a main challenge for VCA remains, and that is the lack of a common processing platform across manufacturers, Bharadwaj said.

“For example, each camera vendor has a different processor type. This requires VCA companies to invest a huge amount of R&D in supporting each manufacturer.”

Servers Making a Comeback

●Zvika Ashani, CTO of Agent Video Intelligence

VCA on edge devices is growing in terms of market size, but so is server-based intelligence, said Zvika Ashani, CTO of Agent Video Intelligence. “Due to processingpower limitations, VCA which resides completely inside the edge device is mostly limited to relatively simple applications, such as tampering detection, people counting and directional movement detection.”

“When you need to perform analytics or correlate metadata on live or stored video from hundreds of cameras, you need the greater processing power and central management capabilities of a serverbased solution,” Bohn said. “Serverbased analytics enable more complex analytics and fast searches through archived video. Server solutions are also critical for making a variety of video analytics tools available to various departments, from security to operations and marketing, and allowing for searches of after-the-fact analyses, such as facial recognition, that are beyond the capabilities of on-camera solutions.”

A future-proof solution is to enable video analytics at both ends — server and edge, Bohn continued. “The need for a strong video management system with a unified yet flexible integration framework will also become more evident, as the value of querying across metadata from different edge devices from multiple vendors will be very high.”

A surveillance system that integrates information from multiple VCA-enabled cameras to a central server does offer a better picture of the entire monitored area, Lim agreed. “The system can take preemptive actions, such as triggering brighter lights where the intruder is detected or notify security to intercept the escape.”

There should be no debate of where VCA should reside, Troha added. “It all depends on how value can be delivered to the end user. VCA is only an ingredient in the video system and its purpose is to help the end user extract the most value from their investment in advanced surveillance.”

The end goal is to employ technologies that give the customer an optimal solution, Bharadwaj said. “VCA in itself is not a ‘be all, end all' solution; to maximize overall system reliability, even analytics solutions that are not video-based need to be integrated.”