People counter or footfall counter that helps the retailers count the number of visitors is being used in the fashion industry since quite a significant time now. From infrared beam to video analytics, the footfall counting technology in fashion retail over a period of time has seen huge transformation. Today the most advanced level of this technology is not just helping the retailers count the footfall, but also supports in analysing the consumer behaviour in retail stores. This article explores how video analytics technology is helping retailers avail a number of benefits which assist them with better retail analytic capabilities.
Closed Circuit Television (CCTV) cameras have been used by retailers since long. These are used for preventing shoplifting, reducing employee theft, safety etc. Today these CCTV cameras, equipped with advanced level of technology like Wi-Fi and AI, are helping the retailers to keep an account of the footfall in the stores. The advancement in technology also provides some added advantage over counting the number of visitors like consumer buying behaviour, their movement in the store, time spent at a particular section. The data collected through this is used by the retailers to take better informed data-driven decisions.
There are different kind of technologies for counting people like infrared beam, thermal counter, video and Wi-Fi analytics. The most advanced level of technology uses technologies like Facial recognition, AI, ML and computer vision along with CCTV to deliver the dynamic need of the retailers. Moreover, each advanced level of technology has overcome the flaws and inaccuracies presented by its predecessor. Also, with the expansion in the retail formats, there has been a need for a system with better accuracy rate.
The advanced level of technology is designed to deliver better accuracy rate when compared to the other level of technology available. The technology also can bifurcate the counting based on different parameters like age, gender and can also differentiate between the staff and the actual visitors of the store.
There are different technology providers which are using video analytics with different technologies to provide the retailers with best of the solution. Below are few of the leading providers that provide the retail industry including fashion with the most advanced level of technology available in footfall counting.
ShopR360: CCTV cameras and Wi-Fi technology
ShopR360 is a well-known technology provider known for providing footfall counting technology to retail, mall and hypermarket. The company came up with the solution to help the offline retailers with data that would help them with better retail strategies. “We wanted to create a technology that would protect the privacy of the customers and will not capture any personal data; also one that would not require any new hardware and equipment for installation,” said Pranav Bhruguwar, Founder and CEO, ShopR360.
The technology uses the existing cameras installed in the stores to analyse the footfall count in the store. One of the advantage is that the technology not just helps in counting the number of people, but can also analyse the customer behaviour inside the store. This is done through Wi-Fi technology that uses the MAC Id of the smart phones of the customers to come out with data.
The Wi-Fi technology works as the smartphones are designed to continuously search signals in the environment to latch on to a network. “These MAC Ids tells nothing about the person but reveal only the device information. Once the MAC is sensed from the device, on the basis of signal strength and time stamps, it can then tell how much time a customer spent in a particular area, movement of the customer in the store etc,” informed Pranav.
ShopR360 is much concerned about privacy, and therefore, allows special benefits. All the data interpretation is done in the store environment rather than on cloud. This computed data is then sent to the cloud which is then made available to the retailers on their dashboard. “The whole point is not to capture the data in the first place and then completely anonymise the data. We believe in computing on the edge which means that all the data is computed in the store itself and only the Meta data and anonymised data go on the cloud so that no one can misuse them,” asserted Pranav.
When the CCTV cameras capture the video, it is transferred to Digital Video Recorded (DVR). These DVRs store the video and process the data so that it can be used whenever required and this data is then analysed using an open software called open CV (Computer Vision), which is an open source technology that allows analysis of video file. Every form is analysed as a digital blob and each blob has certain parameters like space, colour, form and time stamp which are captured. Through this CV algorithm, the retailers are able to identify where is the digital block moving, what is the form of the block person, dog and object.
“What happens is that accuracy with open CV is only 60-70 per cent, so we create a layer of machine learning and AI on top of this to then improve the accuracy,” shared Pranav. Above this open source, the company has established its own algorithms using AI and machine learning that helps it identify the customers and their behaviour increasing the accuracy percentage to more than 90. The technology can identify people based on number, gender and age according to the need of the retailers.
For identifying the difference between the staff and the customers, the company has built certain parameters like uniform, tag, batch etc., that would differentiate the staff of the store from the customers. “Most of the staff in the retail store have some kind of identifiers with them like uniform, tag, gear, and this along with AI, monitor the same digital ID in the store moving in different places for a longer duration; this can help eliminate that as staff or sundry movements,” told Pranav.
Klik Analytics: CCTV cameras and facial recognition technology
Started in 2017, the company is a Bengaluru-based start-up that provides footfall counting technology with facial recognition technology. The company initially worked on creating overhead footfall counter for retailers. These were the basic level of footfall counters and only counted the number of people entering and leaving the store. The solution made about 250 installations in the retail space. However, the feedback from these retailers, revealed the need for a more advanced level of solution that could differentiate between the customers and employees along with the gender and age information.
This new need led them to the creation of an advanced level of solution with facial recognition technology using the CCTV cameras that are already installed in the stores and malls. klik Analytic footfall counters, built with technologies like Artificial Intelligence and Data Analytics, help in deeper analysis of customer activity in the store. The features include real time footfall tracking, staff management, peak hour’s data, conversion rate etc.
The CCTV cameras used for installations have to be IP cameras that are installed in the store along with a device that the company provides. The technology works by taking the CCTV video footage from the camera and processing the footage through their embedded board, a small device that needs to be installed in the store. The device works by analysing the video and extracting different parameters like face of the customers. The algorithm designed then analyses the video to identify age and gender of the customers. The data collected here is then sent to the cloud where data analytics plays its role to calculate the number of stores, gender, age that are delivered to the retailers on a dashboard. “The hardware required is an IP camera besides internet connection along with our product that comes with own embedded board that processes the information,” informed the company spokesperson.
One of the major concerns while implementing any such technology is to enable it to differentiate between the employees and the customers in the store. klik Analytic Facial Recognition feature can store the staff data or photo within their system and identify them easily. The final footfall data that is generated then can easily subtract this information from the overall data that the system will record. “We feed the data and images of the employees in the system to differentiate them from the customers. So, basically this is an advanced level of footfall counter and only a few players are actually working on them,” stated the spokesperson, further adding, “The proprietary algorithm that we have developed is an open source tool as per which we have developed our facial recognition model and we have accelerated it to make it work on every single hardware.”
The technology can deliver an accuracy rate of about 90 to 92 per cent and the company is still making efforts and aims to increase this rate further. “As a part of our future plan, we are working on to increase the performance limitation. For example, GIP is one of the malls that we have worked with. It has multiple entrances and hence, it becomes more challenging for us to track customers entering through different entrances; this is the sort of challenge that we are looking at. The journey that the customer takes up within the shopping centre is that we are looking into. Going ahead, the path that the common customer will take up is the next thing we are working on,” informed the company spokesperson.
Capillary: AI–powered footfall counters
Based in Bengaluru, Capillary Technology VisitorMetrix is an AI-powered footfall counter for offline retailers. The technology uses cutting-edge AI technology based on computer vision and machine learning-powered visitor counter that allows offline retailers to get insights into their walk-in visitors’ behaviour.
The technology helps retailers with powerful retail insights around customer demographics, store conversion and the impact of marketing campaign. The intelligent people tracking system captures accurate age and gender demographics data to help one personalise customer experience. Moreover, the retailers are benefited with a dashboard that helps to track in-store conversion ratio and power house where most conversion take place. It is designed to compare real time and historical traffic trends across store and regions to identify locations and the best conversion rate.
The algorithm designed is trained with more than 3 million images to identify humans from the top view. Moreover, 9 million impressions of non-human objects such as shopping carts, bottles, helmets etc,. have been used to train the algorithm to differentiate between human and non-human objects. Further rounds of intense image training have also been followed to make the algorithms capable of identifying visitors to the store accurately irrespective of store conditions.
The intensive training of the algorithm has enabled it to deliver an accuracy rate of about 95 per cent, which continues to increase with time and power of processing more number of images.
The technology assists brands to analyse historical store traffic patterns to determine staffing requirements but can also give immediate responsive data to changes, including studying the impact of promotional campaigns. Brands can use this information to strategise their operations and marketing campaigns to look deeper at their performance and pull the right levers for improved productivity.
The advanced level footfall counting technology helps the retailers with a number of benefits including boost in sales and conversion; it can also help retailers to increase their sales and conversion through a combination of several factors like improving store layouts, understanding customer preferences based on dwell time, personalised product offering based on customer age and gender and effective staff allocation around store peak offer. This also helps analyse the effectiveness of marketing campaigns by monitoring spike in footfall, dwell time near specific product and overall conversion ratio for a promoted product.
Evolution of People Counter
The realisation for the need to have a people counting technology in retail stores led to the discovery of various ways of counting people in the store. Manual clickers are said to be the very first technology which required staff to stand at the entrance and the counting device was then clicked each time a customer entered the store. After that, pressure sensitive mats were used that allowed counting walk-ins using pressure sensitive sensors placed at the entrance of the shop.
With the advancement in technology, the Infrared beam counters came up which are devices located at the store entrance that continuously emit beams and each time the beam is broken, a count is recorded. These were soon considered inaccurate for the variations they showed when multiple visitors enter the store, and their inability to differentiate people entering or exiting the store.
Thereafter, Thermal Counters were the next to come up which worked using heat sources. One of the challenges that this technology had is its susceptibility to weather conditions and humidity. Also, these provided low demographic accuracy.
Video and Wi-Fi counters were therefore invented to capture the changing requirement of the retailers using complex algorithms and image processing, and thus offered higher accurate footfall analytics and customer demographic data. Wi-Fi people counters use Wi-Fi probe request signals from customer smartphones. The limitation with this technology is that the accuracy is compromised by shadow and low lightning and inability to distinguish between people and object.
Nowadays, computer vision and AI-powered people counter are in use which offer in-depth customer insights like facial recognition, speech recognition and even map the previous buying behaviour. They also browse data to predict the next buying behaviour. These are considered to be most accurate as they possess the ability to work under low lightning conditions and can clearly distinguish between people and object.