To start with, can you briefly define what intelligent video is?

- Video intelligence or video content analysis is the ability to mathematically detect and handle events using video cameras.

What can intelligent video accurately achieve today?

- Today, video intelligence can reliably detect intruders, license plate numbers, illegally parked cars, loitering people, crowds, unattended bags and some other types of events. Video intelligence today cannot yet reliably recognise faces in crowded areas for example.

When talking about intelligent video there seems to be a lot of concern about false alarm rates, is this part of the reason why intelligent video hasn't really become properly established yet?

- False alarm rate is definitely one of the major issues in the video intelligence market. Only a few products deliver low false alarm rates, especially in challenging environments like railway stations, city centres etc. We are one of those, Agent Vi's IPoIP (image processing over IP network) solution has been successfully deployed, and numerous customers are experiencing very low false alarm rates.

For example, the Israeli railway uses Agent Vi's solution to protect train stations in Israel thanks to the very low false alarm rate; this was concluded after a thorough testing period of 18 months.

What is the main difference between video motion detection and video analytics?

- Video motion detection is only one type of video analytics, video analytics is a much wider term that is used to define many kind of events which may be related to moving objects, or non moving objects.

Will intelligent systems be able to replace CCTV operators completely in the future?

- Video intelligence can definitely replace a lot of the functions that today are done by humans. As video intelligence solutions mature they will be able to replace more and more functions currently handled by human beings. Operators cannot concentrate on CCTV monitors efficiently for longer periods of time. Researchers indicate that the human ability to detect events decreases by 90 % after 20 minutes of watching monitors.

Intelligent video can be complex, how can companies within this market help customers get a better understanding?

- The challenge of video intelligence manufacturers is two-fold, providing reliable solutions on the one hand and very simple user interfaces on the other. We have invested huge effort into enabling our customers to define very complex event detection rules in a very simple way. In our case this is done by a special wizard that guides the user step by step on how to define all the rule perimeters.

What are the major limiting factors for the intelligent video industry?

- In order to achieve higher market penetration of video analytics the first thing that has to be achieved is higher reliability of existing products. Many customers are still sceptic about the reliability of video analytics systems; another issue is the vendors' ability to respond to new customer requirements for new analytic capabilities in a short time, today the development cycle of new analytic products is relatively long. Agent Vi is using IPoIP technology and most of the new algorithms are developed on the VI server side and not in an embedded environment, therefore our development cycle of new products is relatively very short.

Have there been any initiatives to directly link triggered events, for example, reporting illegal graffiti painting in a public setting to local police authorities?

- Most of the video intelligence systems today are used by the public sector; these customers usually control all their cameras from central control stations. Video intelligence systems automatically detect intruders, license plate numbers, unattended bags and other events and send an alarm to an operator in the central control room.

Do you see any other applications, other than surveillance, for video content analysis?

- Absolutely, two important examples of other vertical markets using video intelligence market are the retail market and intelligent traffic systems. For example a retailer can obtain information about the customer flow, hot-spots in the shop, the length of the queue automatically by using cameras which in most cases are already installed in the stores. Toll roads can automatically receive information about vehicle flow and any traffic incidents. So these two examples show that intelligent video can generate more revenue and improve an organisation's efficiency within other industries not just in the field of security. Security is definitely not the only market for intelligent video.

What do you think video content analysis will mean financially for end-users in a long-term perspective?

- Video intelligence saves money. Video intelligence can save manpower, shrinkage in stores, loss of revenue in toll roads because of traffic obstacles and many other applications. Today cameras are not used efficiently and are not saving any expenses for customers compared to intelligent cameras.

The key issue is the scalability and the ability to use existing equipment; we are offering our solution today using existing cameras such as Axis IP-cameras. Therefore the customer can upgrade regular cameras to become intelligent without purchasing any new equipment. The software update is done remotely; avoiding the need to send a technician to carry out the system upgrade.

Explain the advantage of your IPoIP-technology?

- Video intelligence is based on image processing technology which has been known for more than 30 years, the major problem for video analytics systems is to deploy high levels of algorithms, mainly because of limited processing resources. IPoIP technology is a unique concept of distributing the algorithm into two separate components, one is called Vi-Agent, and this software component is installed in or near the camera and carries out the initial processing of the video. The results of the Vi-Agent are then transmitted over the network to the other component of the system which is called Vi-Server. The Vi-Server analyses the data and completes the detection algorithm, by making this distribution, which is unique to Agent Vi, we solve the limited processing resources problem and enable the use of any level of algorithms.

IMS Research recently released material indicating that the video content analysis market would grow to an estimated total of 600 or more million US dollars and with a growth rate of 80-100 percent by 2010. How a big a slice of this pie do you think Agent Vi will have and why?

- Agent Vi's offering is designed for enterprise grade customers, that includes customers which are using hundreds or even thousands of cameras. I don't want to mention any numbers but we are already one of the major players. The intelligent video market is very young and there is no clear leader today.

Where do you see the intelligent video market in 10 years?

- I believe that in 10 years every video surveillance camera on earth will have some kind of intelligence, some of them will have basic capabilities like motion detection and some others will have very sophisticated capabilities, like recognising suspect faces in crowded environments and automatically tracking them when moving between cameras.