July | 2020
Large-scale retailing with the ability to purchase and sell high quantity of products leads directly to high turnover and a safe operating margin for the retailer. Lateral spread of margin risks across categories of products influenced by customer and market behavior leads to this outcome.
However, while striving to achieve efficient large-scale retailing, all large retailers face some common underlying challenges related to improving the efficiency of their supply chain operations, and inventory management and pricing (including discount and sales). Compliance (both brand and statutory) is a multiplication factor for the above three parameters and hence becomes the fourth most important factor in large-scale retailing.
AI-driven computer vision technology helps large-scale retailers in automating their entire processes from supply chain management to in-store operations including security and surveillance while reducing human intervention, which are prone to errors.
Addressing retailing challenges with computer vision
Key factors that define the success of retailers is the ability to handle five key challenges
Technology intervention in the last decade has helped customers choose both web-store as well as brick and mortar store. This has given rise to one of the largest supply chain and inventory management nightmare for retailers.
An accurate and on demand audit of inventory in warehouse leveraging computer vision technology helps in managing an efficient ratio of demand and supply while reducing losses due to wastages, overstocking, and on-time availability of products.
Computer vision technology also assists in-store executives and management with repeatability and retraceability (auditability) capabilities, and plugs the gaps in retail store operations while enhancing the customer experience.
Warehouse and stores are highly prone to security issues, which can be efficiently mitigated by having an artificial intelligence driven all-seeing eye that can alert and avoid any safety and security challenges.
Customer loyalty management and compliance to retail industry specific laws will lead to repeat business and industry-wide recognition of the organization. (See Figure 1).
Figure 1: Overview of digital retail operations using computer vision
Inventory and warehouse space management
Indoor micro drone based visual audit can be implemented for stocktaking and space management. Drones would be equipped with autonomous visual navigation with onboard computer for movement between the racks. Drones can do an efficient and error-free stock-take using “ArUco” markers on the pallets and carton boxes.
This helps in frequent stock check to give reverse feedback to manufacturing unit / Inventory procurement unit to ensure adequate stocks are available for retail sales. This also helps in bringing transparency by enabling on-demand audit with absolute traceability, repeatability and log of the visual computer vision based inspection.
It is important to first ship the products with earlier manufacturing date to retail stores/customers in order to maintain a higher shelf life of products in stores. This reduces products being left out in warehouse beyond the expiry date and reduces loss due to discount being offered in retail stores due to lesser shelf life.
Identification of available space across all racks including the ones at the highest row in the warehouse is essential to make sure stocks are well distributed across the warehouse. Along with the above two solutions i.e Frequent and on demand stock audit and First-in and first-out, it would also be highly advantageous to stock products with longer shelf life on higher rows and products with lower shelf life at lower rows in the warehouse, enabling faster circulation of products based on expiry date and better space management.
In-store operations
Plotting a heat map of the store based on the total number of footfalls along with the total amount of time spent near the aisle helps identify the product to be displayed with the possible related combination of another product.
An example for identifying cross merchandising would be promoting batteries with electronic appliances and vice versa based on which aisle is getting more footfall thus increasing the chances of proportional sale of the related product. It is also possible that two totally unrelated items can be cross merchandised based on the customer’s time spending pattern next to certain aisles.
One of the best ways for driving sales and better brand promotion is by placing advertisements and discount banners closer to the areas with highest customer footfall and along the path of customer movement. This is achieved through a balanced insight by combining the results of computer vision customer heat map analysis and customers’ in-store path. The intersection of high customer footfall and the path to the high footfall aisle are ideal locations for brand promotion and sales promotion.
Customer heat map generation can help in identifying the correct placement and rotation of products across aisles, thus enabling a feedback on the planogram adopted by the store. The best way to check if your planogram is working is to ensure the footfall analysis across all aisles is available and see if there is a reshuffle of product categories required to replenish your customer footfall across the entire store.
Studies have shown customers get highly dissatisfied when they are informed that the discount mentioned on the rack where product was placed is higher than what is available on the point of sale (POS) system.
QR code based discount validation along with product identification using computer vision to cross validate the discount and price with the central database using hand held device increases the efficiency of in-store executives. This also helps in auditability and re-traceability with proof of validation for each day.
One of the most important tools in retail store operations is to ensure that there is enough rack occupancy in the front row of the aisle in order to give customers a fuller store perception. An empty store perception will lead to the customer losing interest in revisiting the store.
Also, it is an important brand management tool to identify the total number of products of a particular brand displayed against another brand with similar product in the same or nearby aisle. Computer vision based solution can help give correct visibility and perception.
Many carts are abandoned in the trial room due to apparel’s unsuitable size, fit or color. The customer doesn’t want to return to the fitting/trial room due to queues.
Innovative solution for checking option availability in the store for the chosen apparel by the customer can increase the possibility of sale. Smart apparel assist solution using bar code scanners in the trial room synchronized with the inventory of the store, with an immediate messaging system to in-store executive to fetch the required apparel with customer-chosen new fit, size or color to the correct trial room will boost the sale and reduce abandoned carts.
Compliance and loyalty management
Computer vision can enable dress code compliance for in-store executives to give customers the satisfaction of dealing with SMEs.
Computer vision can help achieve constant monitoring for PPE compliance in areas where it is mandatory to maintain protective gear and clothing to ensure safety and compliance for hygiene reasons and industry and regulatory standards.
Identify promotional displays, floor stands, as per the agreed upon terms with each brand and ensure there is compliance.
When a store is not adequately merchandised, you lose opportunity to improve adequate sales. It leads to customers not being engaged and motivated to buy. Compliance to in-store merchandising has the highest impact on sales lift program. Computer vision can help in ensuring there is information available on near real time basis on the in-store merchandising compliance across various stores, cities and geographies in a single digital cockpit for the management.
Customers loyalty program can be managed using computer vision based face identification to enable digital authentication rather than physical cards or phone OTP. This will enable a better customer experience and future ready program.
Security and surveillance
Identification of crowd at discount aisles and POS using computer vision will enable better store and customer management, thus increasing the customer experience.
Integrated 24x7 fire and smoke detection using computer vision enables better safety and security for your warehouse and store.
Tailgating and unauthorized entry can be identified by using computer vision technology at access-controlled areas by comparing the number of swipes with the number of people who have entered. Similarly, face detection technology can be leveraged to identify persons entering restricted areas.
Digital retail operations with computer vision
Retail operations when streamlined with computer vision technology can help reduce the cost of operations while significantly driving the parameters influencing profit margins.
Narayan Rajashekar
Core member of Wipro’s CTO with nearly 2 decades of IT-driven industry automation experience and specialization in Computer Vision product conceptualization and delivery.
He is also the head of the Go-to-Market team of Wipro’s Computer Vision products and engages with Wipro’s customers as an advisor to identify their needs and tailor-make solutions to automate their operations.
Wars have forced immense advances in medical care. Mobile x-ray machines and sanitary napkins were a result of World War I. Metal plates to help heal fractures were a result of World War II. The use of frozen blood products came with the Vietnam War. Advances in physics, mathematics, computer sciences, radar, infrared, prosthetics, and organ replacement have been propelled by wars.
Communication Service Providers (CSPs) the world over are seeing a sharp decline in connectivity-based voice, data, and messaging service revenues.
Highly advanced and sophisticated robots will function in complex environments and trigger a faster evolution of healthy man-machine collaboration
