Retail sets sights on AI and Computer Vision

Screen Shot 2017-07-25 at 10.58.51 AM



Walmart wants to find out exactly how happy or unhappy customers are with new facial recognition technology. The superstore has filed a patent that uses video camera instore technology to monitor customers facial expressions and movements as they queue. If an unhappy shopper is detected it will notify employees to report to a till, hopefully relieving the shopper of stress.The patent comes as an attempt by the retailer to more efficiently address customer service problems before losing the customer to a competitor. Interesting idea, although we aren’t sure exactly how effective this technology will be. In our experience it seems nobody is ever particularly happy when having to queue?

The superstore previously tested facial recognition technology in 2015 to prevent shoplifters but abandoned the programme as it was ineffective.



At the same time Conduent, a business services company has been awarded a patent for “Facial Expression Recognition Technology”. The technology applies machine learning and pattern recognition to classify images from low resolution cameras. Similar to Walmart’s patent, the technology locates faces in video streams and identifies a person’s eyes to better interpret the meaning of a persons expression. Conduent sees the technology being used in retail and transportation for gauging shopper and rider satisfaction.


Smashbox Cosmetics is using eye-tracking technology to drive usage, try ons and customer visits to their website. In partnership with ModiFace the beauty company has created the MAKEUP iOS app which features eye-tracking technology, designed to provide a deeper level of insight into what users are thinking by measuring what they are looking at. Using eye gaze tracking the app automatically recognises what users are interested in and virtually apply the shade. 

instore technology


Microsoft, Google and more

There are a number of other companies developing facial recognition technology including Microsoft Azure and Google’s FaceNet.


Cloverleaf, a retail technology company announced shelfPoint earlier this year, an AI solution for store shelves capable of using shopper facial recognition to measure emotion at the moment of a purchase decision.


Scientists at MIT are now using Wi-Fi and AI to read human emotions. Using an algorithm that detects and measures individual heartbeats by bouncing RF signals off people. The technology could be paired with Alexa and other smart home devices automatically mimicking music to your emotions.


MIT Computer Science and Artificial Intelligence researchers are using the same AI powered computer vision to identify food. The researchers have trained an artificial intelligence system called Pic2Recipe to suggest recipes based on food photos. Pic2Recipe uses a neural network to find patterns and connections between images and recipes, then predicts ingredients and suggests similar recipes. The CSAIL team is using existing datasets to aid their work in addition to combing websites such as All Recipes and to develop their own database. Using a photo of a food item, Pic2Recipe is able to identify basic ingredients such as flour and eggs then suggest similar recipes based off similar images in the database. The findings are part of a paper that will be presented later this month.


Related Items