| | August 20198ideo 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: -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 VIN MY OPINIONBy Sanjay Bakshi, Sr. Program Manager (Database Services), Safexpress Private LimitedVIDEO ANALYTICS AND ITS RELEVANCE TODAYSanjay Bakshi
< Page 7 | Page 9 >