If you sat down with a person who had a bird’s eye view of the entire store and asked them to track every movement (person or object) for the whole day, you would get the manual version of computer vision. Computer vision essentially allows technology to possess “eyes” and make use of the data that it analyzes, adapting them to real-life scenarios. Let’s dive deeper into what Computer vision entails, its various applications, what evident role they have in the retail industry now and in the near future:
- What is computer vision?
- Present and future of computer vision
- Computer vision use cases in retail
- Innovating self-checkout
- Inventory and theft management
- Customer journey heatmap
- Valuable consideration
- Post-COVID significance
- Partial transition
- Final thoughts
What is computer vision?
As surprising as it may sound, computer vision was present on a conceptual basis starting from the late 1960s and has been in development since. The aim was to develop artificial intelligence that mimicked the human visual system with an advanced understanding of images and videos. Those visuals would automatically be analyzed, tagged, sorted, and then the compiled data would be put to use in various outlets. All of that is achieved through image annotation — the process of labeling images with data that you want the computer to recognize through deep learning later on.
Present and future of computer vision
Theory aside, how practical and common are computer vision applications now? Do they hold any substantial potential for the future? Back when the early development of computer vision began, the theory was ahead of the execution. There wasn’t access to corresponding technology to turn the concept into reality. Now, deep learning algorithms are all but a theory — they’re present in our everyday life. This includes facial recognition on social media platforms and smartphones or self-driving vehicles. In the near future, computer vision applications are expected to scale up to much larger use cases that we’ll analyze in a bit.
A handful of prominent use cases for computer vision are in the following industries:
That isn’t to say that computer vision won’t expand to more and more industries, innovating them further with state-of-the-art AI solutions.
Computer vision use cases in retail
In the retail industry alone, whether single channeled or omnichannel retailing, there are a bunch of applications worth analyzing. More and more businesses are considering implementing some form of computer vision technology to automate certain processes. Targeted at specific pain points, the technology will improve the shopping experience for customers while saving resources for retailers. The computer vision applications we will cover are already available and functioning on small scales or limited models. Yet, those are expected to be largely scaled within the following years. Specifically, the worldwide computer vision industry is expected to notice a large spike in value up to $51 billion by the year 2026.
When self-checkout became prominent in supermarkets during the early 2000s, it was revolutionary. It may have been difficult to imagine innovating self-checkouts even further up until a few years ago. Typically, shoppers scan the barcode of the item either with a handheld device or by approaching a self-checkout counter at the end of their shopping trip. Computer vision aims to minimize even this process and make it possible for shoppers to no longer scan items — the computer will do that for them by analyzing their path and products chosen, billing them accordingly.
One of the greatest real-life examples of this is Amazon’s Just Walk Out technology. Amazon opened Amazon Go shops exclusively for their staff which changed the game of traditional shopping forever. Shoppers scan their phones upon entering and then simply gather what they need in a basket or cart. Once done, they leave the store — no lines, no waiting for change. The shoppers are automatically billed after they complete their shopping trip. Amazon’s Just Walk Out technology has already branched out to supermarket chains like Whole Foods that plan to open their doors to 2 new locations with this new mode of shopping in 2022.
- Advantages — One of the biggest perks of applying computer vision to create self-checkout stores is the considerable reduction of waiting times. Queues will basically become nonexistent with “Just Walk Out” stores since the concept of waiting in line to pay for products will be eliminated entirely.
- Disadvantages — With this technology, self-checkout won’t be an option any longer, but rather, it will be the norm. That comes with the risk of significantly increasing the unemployment of cashiers and sales clerks. In the U.S. alone, there are nearly 10 million cashiers whose jobs will be at stake. The silver lining is that even stores with computer vision technology will still need employees to assist customers and restock the shelves.
Inventory and theft management
Computer vision applications in retail will transform the way we perceive traditional inventory and stock management. The AI will track products across the entire store with cameras to make a note of what items are moved from the shelves and where they end up. The technology aims to learn enough that it is capable of linking the products with their corresponding shoppers. This means that when a shopper picks up an item from the shelves, the AI will track whether the shopper puts it in their basket and checks out with the item or returns it back to its initial place.
- Advantages — Not only does this significantly automate stock management, minimizing the need to take inventory of products manually, but it innovates retail security. Similar to security tags on clothing items, even everyday items such as cereal and milk will have digital tags. Store owners will be aware of every single product and notified in case it is stolen by smugglers or miscalculated at the checkout by the cashier.
- Disadvantages — This application of computer vision would have been stupendous if it was foolproof. Preliminary models of this technology have not perfected flaws in the detection process. Additionally, failures in the system may go undetected, leading to unnecessary losses in products if the system is not maintained and monitored by on-site specialists frequently.
Customer journey heatmap
Ever wondered what displays in your store are the most attractive to shoppers? What do they go for the most or least? Computer vision applications in retail offer exceptional data and insight for marketers and business owners by receiving cohesive “reports” of customer activity. The AI will be able to map “hot spots” of the store where most shoppers spend their time or products they interact with most. This will give store clerks a better understanding of what departments of the store are capturing more attention than others.
- Advantages — Access to this data will allow retailers to cast well-informed decisions regarding product placements throughout the store. For example, they may decide to put items that aren’t bestsellers in those proven “hot spots” to generate more exposure. Similarly, they may separate products from those hotspots if they are closer together in order to motivate shoppers to take different routes in the store.
- Disadvantages — Most criticisms of this application question whether or not it is a breach of personal privacy. The AI will essentially be tracking shoppers throughout the entirety of the store to then generate information about their shopping manners. Shoppers must be more aware of what data they are intentionally or unintentionally providing retailers within this case.
The strengthening of AI has been deemed as the inevitable future of technology, and that future may be upon us sooner than later. Besides the use cases that we discussed above, there are a couple of other considerations regarding the current status of computer vision applications in retail and their prospects in our fast-paced world.
The pandemic brought computer vision applications in retail into the spotlight. How? While the majority of country and city populations were in lockdown at home with limited opportunities to leave the house, retailers were keeping their shelves stocked with vital amenities. Social distancing and halting the spread of the virus were a top priority, but since grocery store employees were considered essential workers, they could not quarantine in the comfort of their homes. If computer vision technology was a regular part of more stores across the world, more employees could stay at home and limit the chain of infections. With COVID-19 still present in our daily lives, more and more people are seeking the transition to automated processes in retail and other public industries.
A total transition from current methods in retail to AI-centric is unlikely to occur rapidly. To start, the application of computer vision to the retail industry is a costly change. The implementation of computer vision to a business can cost upwards of $10,000 depending on exactly what processes are automated with the technology. Specialists ranked Amazon’s Just Walk Out technology at around $1 million. That is not a figure that most or all retailers are currently prepared to spend.
For that reason, it is expected that the application of computer vision to the retail industry will be gradual. Businesses will begin opting for technology that automates specific pain points that they are struggling with more than others. For example, a store may give more importance to adopting an inventory management system rather than an AI-powered self-checkout.
Out of the hundreds to thousands of viable computer vision applications, merely the few regarding the retail industry are enough to be revolutionary to the traditional shopping experience. If only a few years ago, all of the scenarios we mentioned above were blueprints, now they are a reality that will soon be available on a worldwide scale.
Computer vision technology will transform shopping as we know it. The concept of queues and cashiers may soon become abundant when entirely AI-centric stores become the norm. Similarly, vital retail elements such as inventory management and security will be automated and nearly flawless compared to manual labor. The need for automation and innovation in the retail industry has never been stronger, especially as we enter the post-COVID era. Are you still left with questions? Feel free to drop us a line.