By Atul Rai
Artificial intelligence-powered video analytics has gained traction to the extent that it has become buzzwords of the 21st century. Right from navigation to security, facial recognition, and even vehicle analysis, AI-based technologies have revolutionised our lives and are continuing to amaze us. As per a report, more than a billion cameras were sold last year. Also, Indians sold 30 million CCTV cameras in 2019-20 with a 20% CAGR. Research says that we have more cameras than humans on the earth. So automating CCTV surveillance is certainly beneficial in several ways.
We install cameras broadly for activity monitoring, vehicle monitoring, person monitoring, and object monitoring. And this monitoring is done by humans themselves. The challenge with human monitoring is that we cannot focus on a particular screen for more than 20 seconds or 30 seconds because of various psychological facts. So the problem is that we are dependent on human capabilities for monitoring and human is not even efficient in that. So that is where automation is needed. Camera monitoring is bifurcated into two parts –
1. Loss prevention or direct monitoring
2. Secondary or revenue generation monitoring
Direct monitoring which we see in manufacturing, infrastructure, retail, or government organizations where activities like accident monitoring, violence monitoring, cleaning mopping monitoring, person detection are monitored. Also, facial recognition where we are doing identification whether it is for attendance or to detect blacklisted/whitelisted persons also comes in direct monitoring. Its purpose is to prevent any loss and is a cost-saving method used for monitoring. For example, if an accident happens that comes with monetary loss or any violence will lead to sabotage in store. This is prevented by sending real-time alerts so that necessary actions are taken to avoid any loss.
When it comes to secondary monitoring it is considered revenue generation monitoring which provides inferences from various activities happening in stores. Stats like the number of people who visited the store, particular apparel they were interested in, at what time there was a hike in the store, footfall gender, etc. are gathered for better decision making. Though video analytics started with a cost-saving monitoring method however it is also working for the revenue generation part. In brick and mortar stores, cameras are installed and also security guards are appointed to prevent any kind of intrusion or illegal activity. So one investment is done on camera infrastructure and another on human force to monitor it. As the world population is huge you can’t put humans behind every monitoring to avoid security breaches. This monitoring method needs to be automated. Take an example of the manufacturing sector where safety jacket checks, loading /unloading time, a vehicle coming in/out, pilferage monitoring or we can say the whole manufacturing ecosystem operations and safety issues are solved using AI-powered video analytics.
Let’s take another example from the retail sector. Currently, the competition of retail stores is with e-commerce. Ecommerce is having the luxury of all the data with them. One of the major reasons why e-commerce websites are winning the retail race is because of their ability to track the daily customer inflows, dwell time, and purchase choices through an automated system. They know exactly what to offer to their customers the next time they open their virtual store.
Therefore, some key customer analytics …….