© 2019 Hey Machine Learning

Face Recognition at the Mall

Problem

According to Forbes global retail showing an annual growth of 4%, while the global market is estimated at billions of dollars. The world's largest retailers, such

as Walmart or Amazon, use state-of-the-art technologies everywhere, which allows them not only to maintain competitive positions but also to increase revenue in offline sales.


On this basis, our customer decided to implement one of the most advanced technologies in the work of the shopping center.


The mall decided to abandon the standard loyalty program for customers for several reasons: the issuance and servicing of cards are expensive, customer service with cards lasts longer than usual, and customers lose cards, which negatively affects the buyer's further involvement in making purchases.

Solution

At the cashier area, we installed a camera that records video with payers. On the server, this video is split into frames, and the neural network extracts key points from faces for further identification. Information about the customers in real-time is saved into the database and loads records about his past visits. If the visitor came for the first time, a new customer profile is created.

The purchase information is synchronized with the POS terminal and entered into the customer profile. In this case, the purchaser automatically receives bonuses for the purchase, and he can also use the available bonuses for the current purchase.

‚Äč

All visitors are aware of the new loyalty program. Cardholders through the email letters, and new visitors by signs, which are located at the entrance to the store and in the cash zone.

Results

As a result of the replacement of plastic cards, the queues for payment have decreased, since buyers do not delay cashiers by searching for a loyalty card. Also, innovation was appreciated by customers, who no longer need to carry cards with them


The neural network processes the face at 15 FPS, and recognition accuracy up to 96%.

See more our works