Customer Profile

US Based Retail Chain

‍

Problem Statement

A Global Retail Corporation operates a chain of grocery stores partnered with ATAI to implement cashier-less store checkout. This new way of in-store implementation eliminates the store checkout issues and unpredictable waiting times at checkouts.

In the overall shopping experience, customer dissatisfaction is triggered only by the lengthy and delayed queues at checkouts. Sometimes the waiting time is huge though the lane is shorter due to the high volume purchase of people in that queue. There were scenarios every day at stores where the payment transactions also caused delays. Also, a time when the customer dropped the items purchased by looking at the queue and this queue size is dynamic in nature with time.  This obviously led customers to choose online purchase options or visit alternate stores. The objective of this partnership with ATAI is to use computer vision and an AI-enabled system for a cashier-less retail checkout.

The major challenge to overcome and deal with would-be customer behavior, i.e., there would be a series that enters multiple picks and places. An item picked up might not be placed back on the same shelf by the customer. Also, keeping track of multiple people who enter the store as a group to make sure that only a single bill is charged to them instead of individual ones. The Landscape of the retail store is another challenge that is difficult to come into action since not all stores are on their own hence be to maintain the same area which makes it difficult to use the same plan for all stores. Each product in-store has its own size and dimensions which cannot be standardized. The system should hence be robust enough to be able to overcome these issues and also be able to maintain accuracy throughout by not charging multiple bills to a single customer any customer nor should they miss any items being picked and placed by any customer.

Solution

The System initiates the cycle from the moment a customer enters the store. A unique ID is assigned to a customer throughout the entire cycle till they exit the store. The placement of cameras at each aisle racks and tracks every single movement of the customer’s behavior, including hand gestures right from the hand moving towards an item to the depth of withdrawing their hand to place an item. Our algorithm can detect the hand gestures to the detail which enables us to give the right prediction of actions for proper decision making which is further used for invoicing. The system also connects with the store management software to obtain the cost of the picked item to add to the invoice. All the cameras are connected in a mesh network that enables the tracking of customers seamlessly and also exchanges information to bring out smooth transactions. The system is set up in such a way that it detects the customer leaving the premises and an invoice is generated automatically and sent directly to the registered payment process.

‍

Results

The results of this implementation led to a huge the experience customer in-store experience which cannot be quantifiable. While the experience of the customer is high, the store owners benefited by utilizing the staff for premium services like assisting the differently abled, senior citizens instead of using them for repetitive tasks such as billing. Beyond all the statistical humans untouched data is available for store owners about the customers’ time spent, purchase patterns, etc. for further predictive analytics

‍

‍