BRS Labs is a software development company that has created the industry’s first behavioural analysis system for video surveillance that adaptively learns behaviour patterns in complex environments.
When Video Analytics was introduced it was positioned as the next generation product to succeed Video Motion Detection – in this respect many would agree that it has succeeded.
However, Video Analytics has frequently been ‘oversold’ over the past few years and generally ‘under’ delivered relative to customer expectations, thus giving Video Analytics bad press and the perception of it being a ‘flawed concept’. ‘Rules-Based’ Video Analytics was always going to struggle to meet the expectation of the customer, because it relies on a human setting up those ‘rules’ (loitering, trip wire, etc). It is not possible to know in advance every potential issue or threat which may exist within a video scene and therefore very hard to accurately set up a rule to detect it.
What was needed was a fresh approach to this challenge and the development of a proactive, intuitive platform that by using Artificial Intelligence could learn a video scene and then watch for unusual behaviour – just the same as a security guard would.
Behavioural Recognition Systems (or BRS Labs) came at this challenge from a totally new perspective, unburdened by the legacy Video Analytic mindset and subsequently invented a bespoke computer ‘language’ utilising Computer Vision technology to give computers the gift of ‘sight’ – a first in video surveillance. The BRS Labs AISight (pronounced ‘eyesight’) system has a unique language, converting regular IP video into Machine Readable Data – now that the computer can literally see, it can make decisions about what it sees within the video stream it is watching. Add to this the gift of real Machine Learning and a sensory, medium term and long term memory, you have a very powerful and unique tool to learn, analyse and report on CCTV video events.
For the first time Artificial Neural Networks are used to analyse raw video data, watching the video scene for unusual or anomalous events in much the same way as a human would. What this means is that the system will learn autonomously the video scene it is watching, understanding what is normal for that video scene and what is not normal – any unusual events are brought to the attention of the operator in real time as a short video clip. There are no rules to set up, no rules to maintain and the longer the system is deployed, the more accurate it becomes, learning time of day and day of week, all with no human intervention.
Video Analytics has evolved into Behavioural Analytics – this is the future………………and the present!