Video surveillance has been around for many years now, however a major blocker to its widespread and effective adoption has been the associated volumes of data coupled with huge computing power requirements leading to its inherent inefficiency.

Imagine a typical security room scenario with a guard assigned to observe the live feed from various cameras in the hope of capturing any exceptional event. He/she is expected to spend all the time in staring at the video feed in the hope of capturing something that merits attention. Additionally, this dependency on the guard to identify an exceptional event brings in the human fallibility in terms of varying focus levels, ability to comprehend, education, training, maturity levels etc.

As estimated by some industry experts about 90% of video feed is never even reviewed.

Hence, though Video surveillance has had tremendous potential; relatively speaking till now, it’s been mostly ineffective.

A famous phrase "Necessity is the mother of all invention" comes to mind in this context since Video Analytics and a plethora of necessary associated technologies, are coming up at exactly the right time!

With major strides in availability of computational power and perhaps more importantly access to open source frameworks like Convolutional Neural Networks (CNN), YOLO (You Only Look Once) object detection model, crowd counting algorithms, AI/ML, Deep Learning etc. Many use cases which were (till a few years back) stuff of science fiction movies only, are fast becoming reality today!

Video Analytics is coming up as the magic pill which can convert the humungous amounts of video surveillance data into real time intelligent and actionable information.

As of today, use cases in the following fields are quite relevant and this is by no means an exhaustive list: -

"Video Analytics could help immensely in terms of classifying Vehicles and aid not only in monitoring but also controlling traffic"

Motion detection -

Could form a critical part of the security implementation in places where movement at specific times and in specific areas is not generally expected. e.g. banks after business hours, locked homes while occupants are away, museums etc.

The technology available allows one to setup the quantum of movement and even define the area in which the movement should happen before an alert is triggered.

Customer count and behaviour analytics -

Retailers can not only count the customer footfalls, but also identify each customer uniquely as an existing customer or a new one to offer more personalized services.

Behaviour analytics like customer tendency to go straight, left or right on entering the store, dwell time on a particular shelf or product, what kind of signage attracts more customers, discounts or attractive bright colours etc. All such analysis is now feasible; thanks to Video Analytics. This provides invaluable insights related to optimal product placements, staff allocation, peak shop times etc.

Facial recognition -

Though Facial recognition is almost taken for granted nowadays, it provides the unique ability to quickly identify an individual by checking for distinct characteristics against the private or public databases. This helps to quickly locate individuals from among thousands of faces in the crowd. Think about a lost child in a crowded fair or a wanted criminal trying to hide in the crowd at airport/railway station or simply a hotel alerting its staff to the entry of a VIP guest. This is bound to bring a paradigm shift in visitor management and attendance systems as well.

Yet another advanced concept related to facial recognition that is gaining traction is Micro Expression Analysis. Correctly understanding our facial expressions, communicate not only our emotions but also actions which are most likely to follow. Facial expressions like Happy, Sad, Surprised, Angry etc. form the basis of Micro Expression Analysis. Video analytics can be achieved based on data curation, sentiment analysis, and other advanced solutions. This could help a lot in areas of Law-enforcement, controlling crime etc.

Vehicle management -

Video Analytics could help immensely in terms of classifying Vehicles and aid not only in monitoring but also controlling traffic. Cameras near traffic crossings can detect traffic violation in real time (e.g. crossing on red light, going over stop line etc.) and coupled with license plate recognition technologies directly raise a challan to the offending vehicle owner based on the registered address. One of the major benefits of video analytics is to have conclusive, irrefutable proof.

Crowd Management -

You can get a nearly accurate count of people attending an event or present in a specific area at a given point in time. This can help immensely towards planning for better crowd management, anticipating security personnel needs and pro-actively managing big events.

The increasing usage of Video Analytics is also accompanied with the evolution of Edge analytics. Which provide a lot of basic functions on the edge i.e. in the camera hardware itself. Though it means faster analytics and lesser dependency on the software/server for generating actionable results, it does make upgrades more difficult since the hardware needs to be replaced.

The Video Analytics market size is estimated to grow from USD 1.69 Billion in 2016 to USD 4.23 Billion by 2021, at an estimated CAGR of 20.2 percent from 2016 to 2021 (Markets and Markets 2016 report on global analytics market).