Computer Vision Aided Smart Merchandising and Customer Profiling Revolution in Retail

Ever wondered what motivates us to stop, stare, and pick up a certain product at a retail store?

It may be different for different individuals but there is a certain fragrance, sound, color, light that triggers our senses to connect. Retail and CPG giants work extremely hard to get these triggers right to attract more purchases.

Are you a fan of Tom Cruise movies? Can you recall the scenes in Rain Man when Dustin Hoffman could count cards to help Tom play to win? Computer Vision exactly does that.

It is constantly collecting, analyzing patterns in data and helping Retail and CPG to make informed predictions. I think, calling computer vision to be the sixth sense in retail is not an exaggeration.

According to a report published by Deloitte, the top 5 reasons for consumers to select a particular retailer is:

1. Great deals on products
2. Availability of stock at a retail store
3. Convenience to shop
4. Ease of locating product at a store or website
5. Easy check-out process

While there is no silver bullet that can solve all these challenges, Retailers are shifting their focus on adapting computer vision technology to combine consumer videos and understand shopping patterns at stores.

Computer vision allows retailers to observe and unobtrusively make changes to their planograms and promotional offers targeting specific customer segments.

Armored with such a smart mico-observational technology the merchandising teams can have a crystal clear picture of sales per demographic segment. Thus retailers can boost sales and increase their return on investment at a faster pace.

Why Computer Vision is the way-forward for Retail and CPG industry

A market research firm Escalent reported a 3.4% increase in in-store retail sales growth in the United States in 2019. The total revenue of in-store sales was $152.7 Billion versus $62.5 billion of e-commerce.

Precisely this is the reason why retailers are increasingly adopting deep learning (computer vision) models to identify objects, classify, and summarize images that would normally skip human eyes. These CV based observations provide useful insights to increase sales in the brick and mortar set up.

There are a plethora of use cases where deep learning technology is aiding retailers to achieve sales goals, here are the latest ones:

1. Computer vision technology to create heat-maps of a retail store. The heat maps are not only created by monitoring the density of people walking through the aisle but also by capturing the details of the products they touch.
2. Computer vision enabled self check-out. Standing in the payment line, waiting for the person to scan each barcode before we make payment is a tedious experience. CV identifies products and bills customers accordingly. Nowadays the payment is also received using facial recognition technology.
3. Smart Product Information Management system (PIM). Using a smart PIM customers can take the picture of a particular product and CV will assist them by detecting the product and providing necessary information about it.
4. Virtual mirrors incorporating CV and augmented reality is the new type of dressing room. Using this technology users can try on merchandise in various colors and designs without actually trying them.
5. Robot assistants powered by computer vision are proving to be effective sales personals. They interact, guide, and help shoppers in the best way possible. LoweBot and Tally 3.0 are the best examples.

Sky is the limit to highlight the ideas and innovations aided by Computer Vision technology. By narrowing our vision for this blog I intend to highlight the use of Computer Vision to observe, monitor, and capture consumer interactions inside a retail store. Let’s begin!

Computer Vision for Smart Merchandising in Retail

According to Gartner, 85% of customer interactions in the retail industry will be managed by AI.

At this pace, the conventional methods to stock up is not going to work for Retail and CPG folks.

Here are the smarter ways of using deep learning technology to enhance consumer product interaction inside the store.

Auditing Product Placement

We are aware of the fact that majority of consumer’s decisions to purchase happens when they are inside the store. Shelf monitoring using computer vision is the best way to understand the product placement and replenish the out-of-stock in real-time to ensure consumers have a hassle-free experience.

Computer vision using deep neural network detects objects within the shelf and classifies them. For images to be accurately classified they have to be annotated with precision. Choosing the right data annotation platform is the key to feed accurate data for computer vision applications.

Check out the fastest and smartest Data Labeling Platform to build AI/ML/Deep learning model faster — Labellerr

Identifying Trends in Product Placement

Hot and Cold shelves is a term all retailers can relate to. A simple break-down of that is if a product is picked up by maximum customers all the time then that is a hot zone and vice-versa.

Tesco PLC uses image recognition technology to improve product placement in its stores, the result of which is increased sales.

Image recognition photographs are studied in real-time with the help of computer vision technology and it gives insights like:

1. Products most often picked up by customers of that particular region
2. Product bundles preferred by customers (beer and diapers is the most popular bundle in the USA)
3. Tracks/ aisle followed by most customers
4. Promotional zones most visited by the customer
5. Least section of consumer interaction on the racks in a store
6. Where do customers spend the most time?

And the list is humongous.

Customer Profiling Using Computer Vision

Gourmet candy retailer in the US, Lolly and Pops uses computer vision (facial recognition) technology to identify and reward frequent buyers.

As the buyer walks in their image is captured by cameras in the store and in real-time an intimation is shared with the person at the billing counter. Loyalty rewards always draw customers more frequently to the store and Lolly & Pops have taped on it quite impressively.

They also monitor customer preferences and share recommendations with them based on their previous purchases at the store. Isn’t that an incredible buying experience for every shopper?

The important fact to be noted is all of this is only possible if the tags are accurate. The face of the customer should be tagged with the right profile in real-time. That can only be accomplished if retailers opt to feed in the data that is accurately annotated.

Labellerr provides high-quality machine learning assisted data annotations with unbeatable service. Take a free trial to experience the magic of the smartest Data Labeling Platform for Retail and CPG. Data Annotation to Build Incredible Computer Vision Applications

The world runs on data and for any machine learning model to perform in the most optimal manner input is the key. The reason why AmazonGo is successful is because of its prompt facial recognition technology. It’s a seamless process for customers to just pick up a product and leave.

For this technology to work successfully the most important component is data labeling. Every face is labeled with its genuine personal profile and thereby it’s an amazing experience for both brand and its customers.

Feed-in the right data and you will experience the magic of machine learning technologies. Check out the self-help data annotation platform by Labellerr. It enables you to label the data all by urself in the most simple way. You can also hire data annotators from their marketplace. Basically, Labellerr is the one-stop-shop for all your data annotation requirements.

Originally published at https://blog.labellerr.com on January 19, 2021.