Self-checkout is a win-win for customers and retailers looking to shorten checkout lines, but a surprising amount of merchandise is lost at self-checkout stations, impacting retailer profits. Fortunately, technology is coming to the rescue. At the 2010 National Retailer Federation Convention, Intel introduced a proof of concept for a solution designed to help reduce this kind of shrink, as well as help speed up the self-checkout process. Today, boards based on 4th generation Intel® Core™ processors from members of the Intel® Internet of Things Solutions Alliance (Intel® IoT Solutions Alliance) make it easier than ever for developers to implement Intel’s idea and ones of their own as part of massive retail transformation underway with the IoT. The self-checkout loss-reduction solution is simple: an intelligent self-checkout station equipped with video and the processing power for video analytics. Such a solution can to provide immediate alerts of errors to customers and warn store management of potential thefts in progress.



Figure 1. Intel PoC of a self-checkout station integrating a video camera and video analytics to reduce loss and speed up self-checkout.

The Causes of Loss at Self Checkout

Not every loss at self checkout is from shoplifters. They’re actually the exception, though a perplexing problem because they tend to be repeat offenders. Most losses come from consumers simply making mistakes. Unscanned items are often unintentional or inadvertent. Nonetheless, according to one source, shoplifting is five times more likely to happen in the self-checkout lane.


Sources of self-checkout problems and losses include the following:

·        Barcode reading problems due to poor printing or other issues

·        Inexperience with the self-checkout equipment and process

·        Customers hiding things in their cart under bags or a coat so they don’t have to scan them

·        Piggy-backing items—scanning one item, but actually carrying two items across the scanner

·        Scanning lower priced items in the place of higher priced ones of similar weight, such as scanning cheap batteries, but putting higher priced ones in the bag at the end

·        Weighing an item such as expensive bulk coffee beans and inputting a produce code for something inexpensive, such as bananas

·        Covering a barcode and running an item across the scanner to make it look to a sales associate as if an item was scanned.

·        Replacing an expensive item’s barcode with a less expensive item’s barcode


Some of these are shoplifting methods and one of the reasons they frequently work is that retailers get fed up with false alerts at self-checkout stations and disable the weight-based security. False alerts happen with light items such as greeting cards that do not register on the scale to show the item was bagged or when a customer places a wallet, purse, keys, or other personal item on the bagging/scale area and the transaction has to pause for an associate’s approval. Another problem is when manufacturers do a promotion bundling items such as shaving cream with a free razor. If the weight database hasn’t been updated for such a short-term promotion, the customer and sales associate have to deal with an alert.


How Adding Video and Video Analytics Helps

In the PoC, Intel demonstrated that video analytics can spot when an item wasn’t scanned to prevent inadvertent customer errors or theft. Video analytics can also detect when people are piggy-backing products or deliberately replacing the barcodes of expensive items with those from cheaper ones. In the PoC, software provided by NCR enabled this feature, including the modifications required to accommodate the video analytics feed from RTS Flexible Systems.


NCR and Fujitsu have been working with StopLift Checkout Vision Systems to devise a similar self-checkout solution. The result is a system that can instantly recognize when an item was not scanned and send an alert with imagery showing the possible deception to an associate. The associate can then show the customer that they are watching and that the item needs to be scanned. The alert could also go to a loss prevention associate who could handle the situation once the customer attempts to leave the store. The system can also detect merchandise left in the shopping cart or bagged outside of the bagging area without scanning. With real-time alerts, the attendant is notified right away before the customer leaves the checkout. In addition, the system can identify an item not meant for purchase placed in the bagging area, such as keys or a reusable shopping bag, preventing an alert and keeping the checkout process going.


StopLift’s ScanItAll web application provides includes a secure web 2.0 interface to view and analyze the actionable incidents it detects. Combining state-of-the-art web video streaming technology with video-to-transaction log synchronization, the ScanItAll™ web application works on any of the major web browsers without the need of installing additional software.


In addition to solving many of the problems in self-checkout today, such video-assisted self-checkout system also provide an important training asset. Retailers can use captured footage to improve associate training on spotting and preventing loss at self-checkout stations.


Develop Your Own (DYO)

Developers who want to develop a self-checkout system that can use ScanItAll or a video analytics solution of their own would do well to base that solution on a board powered by the 4th generation Intel Core processor product family. These processors are designed to drive new opportunities for connected, managed, and secure intelligent systems on the IoT. To learn more about how they improve POS solutions, I suggest an earlier post “New 4th Generation Intel® Core™ Processors Ring Up Big Retail Advantages.” To learn how these processors excel at handling the video surveillance and analysis tasks required for these kinds of self-checkout solutions, check out “Haswell Platforms Will Meet Demand for High Resolution, Intelligent Video Analytics.”


As for boards featuring these processors, there are a lot to choose from. The Fujitsu Extended Lifecycle Mainboard D3222-B (Figure 2) is a Micro ATX board available with a range of 4th generation Intel Core processors and equipped with an Intel® I217-LM GbE Ethernet Controller to provide stable and manageable network connections. The Fujitsu D3222-B supports the latest Intel® Active Management Technology (Intel® AMT 9.0) security features like TPM v.1.2 and Secure Boot, plus enables remote manageability.



Figure 2. Fujitsu D3222-B


Going even smaller, the IBASE IB908 (Figure 3) is a 3.5-inch disk-size SBC based on the latest low-power 4th generation Intel® Core™ U-series processor. Measuring 102mm by 147mm, the board is optimized for applications in POS and digital signage. The IB908 has two DDR3L SO-DIMM sockets with a maximum module capacity of 16GB 1600MHz. Graphics interfaces provided include DVI-I and 24-bit dual channel LVDS displays. The host of versatile connections and expansions include two Gigabit LAN, four USB 2.0, two USB 3.0, four serial ports, two SATA III ports and two Mini PCI-E slots—plenty of connectivity for however you need to configure it.



Figure 3. IBASE IB908


In the Mini-ITX form factor, Advantech offers the AIMB-274 (Figure 4) with triple display capabilities—VGA/DP/HDMI (DP)/LVDS (eDP)—two COM ports, dual LAN, and one PCIe x16 (Gen 3) and two mini PCIe. For security, remote management, and easier software development, this board supports Intel® vPro™ technology, iManager, SUSIAccess, and Embedded Software APIs. 



Figure 4. Advantech AIMB-274


For those looking for a Mini-ITX board developed with reference to the Intel® Intelligent Systems Framework (Intel® ISF) guidelines, the Venture VG-QM87 (Figure 5) meets its specifications for connectivity, management and security and handling data in a consistent and scalable manner. Equipped with the Mobile Intel® QM87 Chipset, the VG-QM87 supports up to 32GB of DDR3L-1600 memory and has digital display connections for eDP/DP, HDMI, DVI (VGA connection is located in the Intel QM87 Chipset) and the ability to drive up to three independent displays concurrently. It’s also loaded with all the connectivity and expansion slots you’d expect in such an advanced system.



Figure 5. Venture VG-QM87



Self-checkout is a growing and important part of the connected store. With video footage keyed to each checkout action, it’s easier to identify what items and consumer behaviors are slowing down the checkout process, as well as spot theft and collect video that can be used as evidence to prosecute shoplifters. Such self-checkout stations can be easily developed using a wide variety of boards from members of the Alliance that are equipped with 4th generation Intel Core processors. To find more of these boards and speed to market your own intelligent self-checkout solution, visit the Alliance’s Solutions Directory.




Learn More

Contact Featured Alliance Members:

Solutions in this blog:

·        Advantech AIMB-274

·        Fujitsu Extended Lifecycle Mainboard D3222-B

·        IBASE IB908

·        Venture VG-QM87

Related topics:

·        Sensing and Analytics - Top Picks (blogs, white papers, and more)

·        Retail - Top Picks (blogs, white papers, and more)

·        Digital Security & Surveillance - Top Picks (blogs, white papers, and more)

Advantech is a Premier member of the Intel® IoT Solutions Alliance. IBASE and Venture are Associate members of the Alliance. Fujitsu is a General member of the Alliance.


Mark Scantlebury

Roving Reporter (Intel Contractor), Intel® IoT Solutions Alliance

Associate Editor, Embedded Innovator magazine