"Loading..."

Enhance in-store shopping experience utilizing image processing and deep learning

O

ut-of-stocks, overstocks and returns are costing retailers a whopping $1.75 trillion per year, a number to which, if you’re not effectively managing your inventory, your business is contributing. Retail out-of-stocks cost the industry billions every year. Studies into the issue consistently reveal an average OOS rate of 8%. In other words, one out of every thirteen items that a customer wants to buy won’t be on the shelf when they’re ready to buy it. The odds of a shopper delaying purchase after encountering an OOS are only 15%, P&G reports; shoppers are much more likely to simply move to a different brand. P&G estimates the direct sales loss from such a rate of OOS is about 4%. But apart from lost sales, the added costs of dealing with OOS are numerous: extra ordering and auditing eats up time and resources; forecasting accuracy plummets; brand loyalty is eroded; promotions lose impact. While finding/ managing a variety of products in a store, there are many difficulties a customer and store manager encounters on daily basis viz. stock unavailability, product misplacement, confusions among discounts and price, product advertisement and re-printing and pasting price tags at right place at the right time; which need to be smartly addressed.

Customers make it possible for us to continue with our business and its purpose

– Shep Hyken
SMART SHELF IS A SOLUTION TO EMPOWER STORE-MANAGERS OPTIMIZE IN-STORE SALES AND PROVIDE ENHANCED CUSTOMER EXPERIENCE

At InnovatorsBay, we have built a smart IoT platform to enable the customers and store managers have/provide a better in-store shopping experience. An elegant composition of ruggedised self-edge displays, high-end sensors, intelligent microprocessors and a scalable backend developed using cutting-edge technology, forms the basis of the solution. The entire solution can be summarised better into two components: Algorithm and Electronic Price Tag (EPT)


To calculate shelf volumetric occupancy, data from sensors and AI vision camera is processed on a single board computer (SBC) using image processing, deep learning based image classifier and sensor thresholds. As the customer picks up a product, system updates the occupancy to the backend server which is then shown on an interactive web application, enabling the store manager to track on-shelf availability in real time, further alerting about products in-demand, incorrectly placed or out of stock.

Electronic Price Tag (EPT) is an LCD display connected with SBC to display the price of the product, discounts and advertisements fetched from the backend server. The content to be displayed can be updated by store manager on the interactive web application and is reflected spontaneously, making the shopping experience more interactive. We piloted this solution to one of the leading refrigerator door manufacturers in the US.