Video analytics settings

Video analytics are used to examine particular data and events in order to increase your security systems  efficiency. It is important when creating an effective system to create accurate VCA settings and to customise the design based on each scenario and security objectives.  In this blog we will examine best practices for different scenarios.

People counting

It is best to use a camera that is pointing directly down and that is mounted high above the counting area (5 meters).

counting1   counting

When a trackable object enters the scene it can take a couple of frames to identify and classify the object. Make sure that the camera is positioned so there is space around the area you want to count people, to help with the classification process. 

We have 2 options when creating the analytics, we can count the number of objects (people) that enter a zone or that cross a line. For people counting we tend to get better results when using a line but we should be aware of where our line is placed otherwise  our result won’t be great. For example, if the line is by a display area is it possible that we will get false readings from staff crossing over the line to set up the display.

We have also mentioned previously that VCA works by analysing the size of moving objects so be aware of the objects size and direction of movement. A vertical mounting position for the camera gets the best results for people counting but it is easier to mistake a suitcase as a person as they are roughly the same size from the view point, so, given the particular scenario, we may consider to position our camera lower – a suit case is rarely as tall as a human.

Perimeter monitoring

Commonly there are 2 goals with perimeters:

1. Someone is close to the perimeter
2. Someone has crossed the perimeter

Setting up a fence boundary with a line along the top of the fence in the VCA seems to be an effective way to detect intruders that have climbed the perimeter, while a zone or line at the base or just outside the fence/perimeter will detect anyone approaching it.

tripwire2  tripwire - zone & line

In cases of highly sensitive sites or areas, you can have parallel lines on each side of the fence; a normal level alarm on the outside to detect someone approaching and a high level alarm on the inside to detect someone who has crossed.

A best practise when setting VCA lines or zones is place them not too close to the edge of the scene as the object may have already crossed the line or zone by the time it has been classified and won’t trigger any alarm.  Always position the cameras in a way to provide space and time for objects to classified.

Perimeter monitoring comes with extra technical considerations due to the nature of intruders. Instead of walking up to fences intruders are more likely to be running so it is important to have a good  FPS and to add a higher object speed boundary.

Object classifications can be configured to ensure unwanted objects are ignored -you can change the classifications so that it does not detect animals such as dogs or rabbits.  This will ensure that alarms are only triggered when a relevant intruder crosses the perimeter such as a person, car, van or lorry.

Since most perimeters are outside we also have the extra issue of weather. The hermeneutic detection tool in Mirasys System Manager allows you to filter out the background noises, such as changing weather or movement of trees and foliage, and reduce false alarms.

There are many other areas where the use of video analytics can improve the efficiency of your surveillance system. When setting VCA, it is important to understand what is needed to be measured and the scope to avoid to set wrong parameters that will give useless data.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s