Retail and Video Analytics
Talk to most people involved in analytics, and they’ll soon cite the retail sector as a prime example of how the technology is being adopted. To hear some speak you could be forgiven for thinking the retail industry is the proverbial goose that lays the golden eggs! The reality is that retailers want analytics, but installers and integrators need to change their thinking if serious about serving this sector.
[dropcap]M[/dropcap]ention the word ‘analytics’ to someone from the security industry, and it’s odds on that they’ll immediately think about IVA, line crossing, object appear/disappear, entry and exit of a secure area, ANPR and so on. They might even consider audio analysis, in as much as detecting screams, gun shots and other unexpected sounds.
There is nothing wrong with this. It’s what the security industry deals with: violations. We use video and audio analytics as a tool to detect a rule violation or a policy breach. Once detected, we can then trigger a relevant response.
Video and audio analysis is, when all is said and done, detection. Okay, it can be smart and it can deal with more than alarm events, but our collective thinking – and the development of the various technologies – is all aimed at one goal.
Of course, the security sector doesn’t ‘own’ the term analytics. In other industries it means something very different. It could even been argued that the security industry’s understanding of the term falls short of its true meaning. We view analytics as a technology that adds to over prime goal: the protection of people, premises and property.
Now, let’s change the original proposition. Mention ‘analytics’ to someone from the security industry, and then through in a reference to retail, and the thinking slightly changes. Now the rules being considered switch to heat mapping, people counting, loitering detection, dwell time, etc..
Such retail analytics can add value to many businesses, as a quick browse of various IVA manufacturers’ literature will attest. Foremost among these will be retail.
The right solution?
Retail is certainly an industry that can benefit from analytics, and it is also one that will invest in analytic solutions. However, that doesn’t mean they’ll bite your hand off if you offer then a camera which also offers people counting! Retailers are less interested in an alarm/action type analytic.
To be of true value, retail analytics must gather relevant information, correlate it in context with the daily retail operations and present that data in a format or report from which the user can gain a commercial benefit.
Many retailers will already have some analytic systems in place, and these will predominantly be data-based. Whether used to analyse stock holding and predicted sales, transaction values, staffing and resource levels or any other elements of the business, additional analytic systems will need to feed data into these.
It is important to think about the depth of data collected by the systems. For example, a system that informs the user that a premises was visited by 675 individuals on a given day is unlikely to be useful if the client wants to carry out data mining to establish the allocation of staffing. The more data that can be collected, the better it will be for advanced added-value analysis. In security, systems are typically optimised and detect and react. Often a scenario is very black and white: a violation either occurs or does not occur. Such simplicity doesn’t always work for business analysis.
IVA adds value to data-based analytics systems, because most require data entry from the retailer. As such they require a higher degree of involvement. What video and audio analytics add is the ability to automate responses to events that might been seen. If a retailer can establish circumstances requiring action by sight alone, then there could be an option for the deployment of IVA.
For example, whilst a retailer might not have much use for line-cross rules, the technology can be used, along with some supplementary algorithms such as loitering or lack of motion, to deliver a queue management solution. If such a system alerted management that additional staff deployments are required, but could also report waiting times, EPOS transactions during the time period, areas of the store that were busiest and the types of products most in demand, the information suddenly has added value.
A camera which simply detects static people and pings off an alert will not be able to interface with other analytic systems, and might even become so irritating due to nuisance activations that staff tend to ignore its alerts.
The retail sector is, amongst others, a prime target for analytics and business intelligence. For installers and integrators it is a sector worthy of further investigation.
However, analytics solutions must collect the right data – and enough of it – to add value when linked with other solutions. Simply offering IVA as an afterthought will not win many contracts!