Skip to main content

VISION

Next in Line: AI Solutions for Retail Environments

queue management

We’ve all been there: two items in your hand and what looks to be a twenty-minute wait for the checkout counter. (At least no one is paying with a check anymore.) Why, you think, with all the technological advancements we’ve achieved, are we still waiting in endless lines at the grocery store! Isn’t there a better way? There may well be. And, it turns out, it can do more in the store than just solve the problem of queue management.

The retail landscape has changed a lot in the past couple of years: Customer expectations have evolved, and so have the tools and technologies available to solve the pain points in a smart and meaningful way—many of them involving AI. Now, AI comes with a lot of opportunities and benefits, but many retailers are not in a position, or staffed with a development team, to develop and build these AI solutions, even if they know what they want to achieve with them.

Fortunately, there’s a whole ecosystem of implementation expertise out there, which is why Ria Cheruvu, AI Software Architect and AI Evangelist at Intel; and Nicole O’Keefe, Senior Product Marketing and Operations Manager at retail spatial intelligence solution provider Pathr.ai, are here to talk about developing AI retail solutions that really count (Video 1).

Video 1. Nicole O’Keefe from Pathr.ai and Ria Cheruvu from Intel talk about AI’s impact on retail operations and the biggest opportunities for AI solutions today. (Source: insight.tech)

What are the biggest challenges for retailers right now?

Nicole O’Keefe: Customers want that seamless checkout experience—short queues, short wait times. If they’re faced with long queues at checkout they can get frustrated, and it may come to a point where they just abandon their carts and leave the store altogether. That’s the last thing retailers want, and not only because cart abandonment leads to lost sales; it can affect customer loyalty too. They’re also concerned about labor shortages and rising labor costs.

All these factors are challenges we’re seeing in today’s retail space. It really presents the need for retailers to create, in particular, an efficient checkout experience for their customers.

How does AI address some of these pain points?

Ria Cheruvu: Artificial intelligence can definitely be helpful in terms of integrating multiple solutions and developing models around intelligent queue management and automated self-checkout. It can identify and provide understanding of the customer experience and integrate that into a system to provide valuable insights across multiple stores and customers. We’re seeing AI being helpful with scaling that process, too, as well as integrating all the different functionalities together.

How can developers successfully build and implement AI retail solutions?

Ria Cheruvu: It can be challenging, both because of the technical limitations of these models and because of the types of use cases that they’re trying to satisfy. Think of AI being able to count the number of items on a shelf, or identifying it when someone picks up an item and puts it into their basket. It can be integrated into things like a smart shopping cart, smart shelves, or smart robots. And actually a lot of times we can leverage models that are off-the-shelf or use technologies to train and build our own models, which does give us a lot of flexibility.

You do then need to bring in elements like privacy and security. I think there’s a really critical conversation to be had around how exactly we incorporate those things into the algorithms—whether that’s blocking individuals’ faces or respecting their privacy with regard to items they’re purchasing, and basically maintaining their anonymity while also extracting the insights needed to continue to improve the algorithms.

But with the number of AI technologies coming out and the improvements in them, it is becoming easier and easier to have those conversations pertaining to the retail space.

Nicole O’Keefe: At Pathr.ai we have a tagline: You can learn a lot from a dot. Every dot is a shopper moving around the floor plan, and that dot has no personally identifiable information attached to it. So retailers can really leverage privacy-preserving insights to make business decisions in a very unbiased way.

“Consumer behavior is shifting rapidly, and those #retailers who will just want to wait and see how it unfolds are going to be left behind. The time to act through #data is now” – Nicole O’Keefe, @Pathr_ai via @insightdottech

How are you working with retailers to implement these AI solutions?

Nicole O’Keefe: One of the ways we can implement AI is through spatial intelligence, which is all about measuring how people move and behave inside physical stores. We leverage a retailer’s existing camera infrastructure to provide insights throughout a store, in particular around checkouts—understanding how long the queue lines are and how long people are waiting—but in general around how the store operations are running. The goal is for those operations to run like a well-oiled machine and to make the experience for the customers as enjoyable as possible.

Retailers also want to reduce their operational costs and improve operational efficiency, and they can do so by leveraging these insights in a very data-driven way. It’s anything from allocating their resources more effectively to reducing unnecessary staffing costs. They can do things like understand how many registers are being used in a day, and if they’re not being used very often maybe it’s an opportunity to turn over that space to the sales floor and add more merchandise there.

Tell us more about Intel’s role in making these applications possible?

Ria Cheruvu: Our teams at Intel are passionate about building out technologies, but also about providing a foundation for our partners, like Pathr.ai, to then take those technologies forward and innovate on top of them. One of the approaches we’ve taken is around the OpenVINO toolkit, which provides a number of different optimizations and options for building and deploying AI models.

I also definitely point our partners to the OpenVINO notebooks GitHub repository, which has a wealth of information regarding how to get started with OpenVINO and how to build these applications. The way that we’ve designed these reference kits, tutorials, and notebooks is for a partner to basically take it, run it, and see the result. Then they can use it as an inspiration or a foundation to check out additional models, try them for their use case, deploy them on the edge devices that they prefer, and really take it on from there.

We’re also looking very closely at the end-to-end stack, and how Intel hardware can help accelerate a lot of the pipelines and large computational requirements that are required for these types of use cases, especially at scale.

What is the Intel partnership like from the Pathr.ai point of view?

Nicole O’Keefe: Intel has been such a valuable partner for Pathr as we’re scaling spatial intelligence in the retail world. We leverage the Intel® CPU-based edge servers, as well as OpenVINO for our computer vision. We’re able to deploy queue insights at scale in a very cost-effective and efficient way, and Intel has been there from the beginning for that.

How can you spread AI success across the entire store?

Ria Cheruvu: There are a number of different ways we can build on existing pipelines. We’re seeing the emergence of really popular and powerful object detection and classification models, but I would say that it even extends beyond that. There are additional models coming in, like pose recognition and activity recognition, that are helping us better understand how individuals are walking through a store and what they’re doing, which adds to the insights we’re able to get.

In addition, we really need to think about these AI models in terms of the preprocessing and the post-processing we do. For example, once we get those detections, what types of information can we extract from them—attributes and specific types of characteristics. What trends can we form from these models as well?

Zooming out from that, there’s a bigger picture in being able to assemble all of these models as part of pipelines—whether that’s validating outputs across a multiple-camera setup or appending the outputs from each pipeline as part of a dashboard for easier visualization.

Where do you see this space headed?

Nicole O’Keefe: It’s one of the most exciting things about working in AI in the retail space—figuring out where you’re headed. As customers we’ll continue to want that seamless experience while shopping. But for retailers it’s all going to be focused around optimizing their store operations. That could look like reducing staffing costs by using real-time alerts and understanding the real-time scenario—when do the checkouts need to be open and closed? And then making really data-driven decisions based on that information. If registers are not being used at that moment, maybe staff can be allocated to other areas of the store.

Another interesting trend, which Ria mentioned earlier, is self-checkout. A lot of retailers today are implementing self-checkouts alongside the more traditional checkouts with staff. Here at Pathr.ai, we’re able to empower retailers with insights around both staff checkouts and self-checkouts, and we’re able to help them understand the difference in performance between the two.

Where is AI in retail going, and what does that look like for developers and retailers?

Ria Cheruvu: When developers are turning to models and algorithms like YOLOv8 for object detection and classification, they’re definitely thinking about the bigger picture. And they’re better identifying how their solutions are fitting in a real-world environment—with all of the challenges and pain points that that can come with—knowing that AI models are still sometimes prone to failures no matter how performant and powerful they are.

In terms of the future and where Intel and our teams see spatial intelligence and the retail space going, we’re using existing types of algorithms, optimizing them, accelerating them, and accomplishing new types. We’re seeing a lot of experiences being transformed by AI, and we’re taking steps towards a point where everyone is comfortable with the way that technology is integrated into our environments.

One final takeaway that I would add is about women in AI, and developers who are female and who are driving areas in leadership—and to definitely continue pushing forward with that. With the democratization of these reference kits, and implementations that you can just plug in and use, I think that’s a very big motivation for being able to get started in the field. That’s something we definitely want to see more of in the AI space.

Nicole O’Keefe: Consumer behavior is shifting rapidly, and those retailers who will just want to wait and see how it unfolds are going to be left behind. The time to act through data is now. And one of the ways that they can stay ahead of the game is by using spatial intelligence with Pathr and with Intel. I think it’s a perfect combination.

Related Content

To learn more about developing retail technology, read The Future of Retail Technology Is Spatial Intelligence, listen to Streamline Retail Checkout with AI-Powered Queue Management, and join the OpenVINO discussion on GitHub to share your experiences. For the latest innovations from Intel and Pathr.ai, follow them on Twitter at @intel and @pathr_ai and LinkedIn at Intel Corporation and Pathr.ai.


 

This article was edited by Erin Noble, copy editor.

About the Author

Christina Cardoza is an Editorial Director for insight.tech. Previously, she was the News Editor of the software development magazine SD Times and IT operations online publication ITOps Times. She received her bachelor’s degree in journalism from Stony Brook University, and has been writing about software development and technology throughout her entire career.

Profile Photo of Christina Cardoza