Network video underpins next generation of business intelligence applications in the UK retail sector, writes Frank Crouwel of installer NW Systems.
IP-based cameras are being deployed increasingly to help UK retailers uncover operational efficiencies. This trend is confirmed in a key finding from an August 2012 CCTV in Retail survey of 700 UK retailers, by the Centre for Retail Research. The Axis Communications-commissioned study found that 58 per cent of UK retailers plan to migrate from analogue-based CCTV to a new network video system, to integrate network cameras with business intelligence (BI) applications. BI integration was the most significant reason for UK retailers to move to network video, or IP Surveillance as it is sometimes called. So what are these retail-specific BI applications? We were interested to see many of them were geared to capturing and understanding in-store customer behaviour better and flagging up areas for potential improvement. Nearly all these applications only work well when retailers deploy network cameras capable of integrating seamlessly with BI applications through use of open platform infrastructure and APIs (Application Programing Interfaces). The ability to run an increasing range of video analytics software in the camera itself, has definitely supported wider demand for a new generation of ‘intelligent’ network cameras.
See: http://www.axis.com/products/video/about_networkvideo/iv/index.htm
Let’s look at some of the BI applications that are being supported by network cameras in retail. The most obvious one is people counting. It used to be carried out by standalone devices but increasingly the CCTV camera which covers the entrance to a shop is being fitted with people counting software to capture the image of the visitor as well as log his entry and exit. The software produces a stream of metadata which can be easily shared, and cross-referenced with sales receipts for the same period. This sort of cross-referencing is a useful way of determining how efficient a given outlet is. If store visitors are rising, but sales at the tills are static or falling, this may point to a sub-optimal store layout, issues with stock volumes or type or some other management issue.
Intelligent network cameras can also hold dwell-time analytics software to help analyse which aisles and displays are working best to attract and retain shoppers. Cameras can also analyse dwell-time alongside images of customers actually picking up products and putting them in their baskets. Again it is valuable to cross-reference dwell-time data with customer action – normally resulting in purchase. However if lots of customers are stopping to look at an offer or display and then walking away this is a fair indication that something in the offer on display does not work for them. Managers can use this BI to investigate further. It is also possible to use images of customers to analyse visitor demographics, specifically collecting age and gender profiles of customers. Network cameras can go further, through integration with facial recognition software, to determine engagement of specific groups with specific digital or point of sale (PoS) displays. This software is increasingly used to establish the number of unique visitors to a store or display. The resulting facts and figures can give added leverage to buyers when they are cuttings deals with suppliers wanting to display their goods in favourable locations.
To support this type of BI, store owners are increasing deploying network cameras fitted with heat mapping analytics. Heat maps determine areas of most activity and highest footfall in the store. The maps can be used to help direct managers’ discussions about the need for better store design to reduce black spots where few customers venture, and also to stimulate customer flow through the store so they naturally travel next to some of the high value displays. Heat maps can be cross-referenced with till receipt data to confirm the effectiveness of new store layouts.
An integrated network video platform can generate real-time alerts when queues exceed pre-defined thresholds. These alerts should then trigger the opening of more tills and the acceleration of stock replenishment cycles – thus addressing the one customer experience that we could all do without. Some smarter queue analytics software can also integrate with footfall data at store entrances so that extra tills can be opened before that fresh rush of new visitors reaches the check-outs.
Much of our discussion thus far has been about intelligent network cameras – essentially uploading video analytics software into network cameras which are acting as compact PCs. However the reality is that the larger the retailer, the more likely that all the data and intelligence that these intelligent cameras generate will be analysed and cross-referenced centrally. Retail analytics dashboards such as RetailNext. It takes data feeds from multiple sources, allowing retailers to gain real-time access to key BI metrics. This centralised view is increasingly being distributed to key operational managers via smart phones and tablets.
That same CCTV in Retail survey found that 63pc of UK retailers expressed strong interest in having remote access to BI information and in-store cameras via their smart mobile devices –to help managers manage effectively even when their work requires them to be away from the ‘coal-face’ for a few hours. It’s great to see network video finding uses in more and more non-security applications. This is a trend that we believe deserves attention as the video surveillance market continues to mature.
