Everyman’s Amazon Go is within reach—and a lot cheaper to implement
The debut of Amazon Go last month in Seattle offered a smart vision for brick-and- mortar stores, and there’s no doubt the pilot will change the way we shop. However, the “just walk out" store, as Amazon calls it, represents the tip of the iceberg of the digitally enhanced shopping experience awaiting shoppers and store assistants. Mobile computer vision can be deployed more simply, cheaply and diversely to optimize the shopper’s experience and retailer’s profits.
Camera Vision Via Smartphones
The advantages of Amazon Go can be had without the pain of ripping out shelves and cost of implementing fixed digital infrastructure that includes scales on every shelf. Instead, camera vision can be enabled through smartphones, robots and drones; or where desirable, wearables. The beauty of this approach is that it is cheaper and augments human workers, rather than replacing them, thus retaining staff expertise.
Customers kitted out with their normal smartphone, for example, can point it at a shelf of produce and instantly locate vegan or gluten-free goods, or whatever kind of specified product they’re seeking/ Some of the other data overlays that augment the reality captured with computer vision and would be useful to shoppers include products that score highest reviews, or suggested product based on a consumer’s personal shopping history.
Amazon Go is a laudable experiment and it’s good to see Amazon validating the power of computer vision coupled with machine learning. An operation that identifies individual customers and their goods, tracks them and deducts the correct amount from their bank account is undeniably smart. But it’s taken millions of dollars and the custom build of a premises — scaling this model to other retail outlets and smaller shops is just not viable.
Costs
Nor is capital expenditure limited to camera and sensors. As retailers learned during the RFID hype days, the major cost is incurred through mounting specialist equipment on shelves and ceilings and cabling digital devices and equipment. The sheer number of cameras providing data in such a real-time, business critical operation cannot be battery operated and will need dedicated, hard-wired bandwidth.
Interestingly, Amazon has opted to deploy this veritable army of cameras, devices and sensors to achieve a single objective: to kill off the queue and let customers pick their goods “and just walk out.” The Amazon Go model assumes the biggest win for customer and retailer is to remove the checkout queue with its hassle of bagging up. Any seasoned shopper, however, knows that the most time is wasted searching for products.
This is especially true if it’s a customer’s first visit to a store, or you’re new in town. In this common situation, a data service that superimposes a product search and find over the camera vision capture is a quick win. It cuts down on the time and hassle of grocery shopping or specialist product selection for the consumer, and speeds throughput and profits for the retailer, too.
Camera vision enables even additional value. For example, rather than fitting every shelf with cameras in order to capture data, a single camera can be mounted on a robot that patrols the aisles at regular intervals, filming shelves and identifying stock gaps to ensure produce is topped up and maximizing the retailer’s sales opportunity. Similarly, camera-fitted drones can be flown over high shelves in giant warehouse outlets for stock checking or answering customer queries.
Camera vision coupled with mobile devices and deep learning is the way to go for retailers and shoppers, and Amazon Go provides an important proof of concept. Happily for mainstream retailers, many other diverse applications are available to optimize shopping and store operations – at a fraction of the price. The Seattle store will wow – but for everyday jobs such as inventory checking and finding the right product, look no further than your smartphone.
Camera Vision Via Smartphones
The advantages of Amazon Go can be had without the pain of ripping out shelves and cost of implementing fixed digital infrastructure that includes scales on every shelf. Instead, camera vision can be enabled through smartphones, robots and drones; or where desirable, wearables. The beauty of this approach is that it is cheaper and augments human workers, rather than replacing them, thus retaining staff expertise.
Customers kitted out with their normal smartphone, for example, can point it at a shelf of produce and instantly locate vegan or gluten-free goods, or whatever kind of specified product they’re seeking/ Some of the other data overlays that augment the reality captured with computer vision and would be useful to shoppers include products that score highest reviews, or suggested product based on a consumer’s personal shopping history.
Amazon Go is a laudable experiment and it’s good to see Amazon validating the power of computer vision coupled with machine learning. An operation that identifies individual customers and their goods, tracks them and deducts the correct amount from their bank account is undeniably smart. But it’s taken millions of dollars and the custom build of a premises — scaling this model to other retail outlets and smaller shops is just not viable.
Costs
Nor is capital expenditure limited to camera and sensors. As retailers learned during the RFID hype days, the major cost is incurred through mounting specialist equipment on shelves and ceilings and cabling digital devices and equipment. The sheer number of cameras providing data in such a real-time, business critical operation cannot be battery operated and will need dedicated, hard-wired bandwidth.
Interestingly, Amazon has opted to deploy this veritable army of cameras, devices and sensors to achieve a single objective: to kill off the queue and let customers pick their goods “and just walk out.” The Amazon Go model assumes the biggest win for customer and retailer is to remove the checkout queue with its hassle of bagging up. Any seasoned shopper, however, knows that the most time is wasted searching for products.
This is especially true if it’s a customer’s first visit to a store, or you’re new in town. In this common situation, a data service that superimposes a product search and find over the camera vision capture is a quick win. It cuts down on the time and hassle of grocery shopping or specialist product selection for the consumer, and speeds throughput and profits for the retailer, too.
Camera vision enables even additional value. For example, rather than fitting every shelf with cameras in order to capture data, a single camera can be mounted on a robot that patrols the aisles at regular intervals, filming shelves and identifying stock gaps to ensure produce is topped up and maximizing the retailer’s sales opportunity. Similarly, camera-fitted drones can be flown over high shelves in giant warehouse outlets for stock checking or answering customer queries.
Camera vision coupled with mobile devices and deep learning is the way to go for retailers and shoppers, and Amazon Go provides an important proof of concept. Happily for mainstream retailers, many other diverse applications are available to optimize shopping and store operations – at a fraction of the price. The Seattle store will wow – but for everyday jobs such as inventory checking and finding the right product, look no further than your smartphone.