Video Analytics Deliberations

The Meeting

On a recent visit to a Security trade show, I met a company that supplies video analytics and posed a few questions. Somewhat surprisingly, they were hesitant to give simple advice and statistics without having a proper awareness of the site, conditions and customer requirements. This is probably indicative of the situation where people are taking and pushing analytics in the attempt to sell themselves and their product, but often with little regard to the environment where it will exist and what is genuinely required.

The Issue

So what precisely is video analytics and in today’s world who needs it and who wants it? Who really cares where this is actually installed?

Wikipedia says that the definition of Video Analytics is, in essence, where a processor & software looks for some specific type of information, or changes in a video stream, for example, colours, patterns, repetitions, etc. Internet search engines are similar in that we need something, and the search engine looks for it amongst massive amounts of available data.

Analytic Choices

With analytics, there is a myriad of variations, such as line crossing, ANPR, traffic assessment, congestion warning, queue management, point of sale overlay, parking management, and the detection of faces, intrusion breaches, abandoned objects, colours, smoke and fire, etc. So, the more important part is to think carefully about the project and what analytics it truly requires.

Some analytics are still unsuitable, emergent or even unknown, which makes the industry exciting with future development prospects. It is vital to balance practicality, costs and accuracy with every project. One would not spend a massive amount of money to detect stationary vehicles in a housing estate. We would not use low accuracy to discover perimeter breaches in crucial business parks, and we would likewise not use very high resource software in video that has many changes and variables. Instead, match what is available to produce a required and reliable result, and start thinking in terms of multiplicity or flexibility of analysis, where we use the different analytics at different stages so that each solution operates in its preferential design setting.

Stages to a Solution

When we use analytics, we begin by clarifying the request or suggestion for a particular video analytic. If there is no basis or purpose for the requested analytic, then it is as worthwhile to an end-user as purchasing dog food but without owning a dog. Once the specific scene with all its environmental conditions are understood, as well as to what the user needs to extract or be notified of specifically, then one could move to propose the brand and type of analytics to consider. Be aware that weather and lighting conditions change, and the precision of the software also impact any given solution.

The final proof would rest in the implementation and successful results. Does it produce the required results? Do a test to document the effectivity of the implemented solution.


‘Plan to Prosper’, have a detailed discussion and understanding of what is needed combined with the support of a projects team from Sensor.

Contact your nearest Sensor Security branch today or get in touch with our customer support department.

Technical support: +27 (0)861 736 787

Email address:

Scroll to Top