 Hello, this is Henry Lau with HP. Thank you for my talk on self-checkout theft detection using EdgeX framework. A little background information to start with. NRF stands for National Retailers Federation. It holds a larger show in retail space every January in New York City. It is attended by over 40,000 industry professionals and over 18,000 retailers. HP has been a participant in this show for many years. For this past January 2020, within our large booth, we have over 15 theme stations covering product introductions, showcases of our partners, and also demonstrations of upcoming ideas and innovations. In one of these stations, we put together a self-checkout theft detection showcase by making use of EdgeX framework and combining four data streams and recognize the events in order to determine theft. Let's look at it this way. This is a typical self-checkout station you see in many grocery stores. Starting from the left, you have the staging space where the items you intend to purchase, maybe on a basket, is put there. In the middle is where there's usually a scanner and then also a touchscreen for you to enter a search code or something like that. And then the right-hand side is the bagging area after you scan the items, and there's usually also a receipt printer attached somewhere. So the way we did it is we don't have enough space in the show to put together a setup exactly like this. So we kind of mock it up, right? So we have put in the HP engage one point of sale system as the touchscreen running one of our partners point of sale software. The scanner is one of the industry standard scanners from DataLogic, and then the bagging area has a weight scale that detects when something is put on there. And also on the top, we mounted a camera to look down so we can watch how items are being scanned during the checkout process. And all these data streams is sent to an edge gateway for central processing. The idea is to combine all these four streams, the point of sale transaction, scanner data, the weight scale data, as well as image recognition coming from the camera, and combine it all, process it on a on-premise gateway that's based on the EdgeX framework. This is a view of the architecture. So from the bottom is where all the devices are coming in. In this case, the point of sale transaction, the scanner scale and the weight scale information all coming into EdgeX via the MQTT protocol. For the USB camera, it is within the gateway is running a object recognition algorithm to detect whether we are scanning a bottle of olive oil or scanning a apple, banana, etc., etc. The result of that object recognition is also fed into the EdgeX framework via the REST API interface. We have configurations within EdgeX on how to process these data streams. Up on the top are the application layers where we can process the events, reconcile the events, do a product lookup, and for the sake of the show, we present the results in a UI dashboard. Basically, we flash a red icon if there's something wrong with the scan, and we put on a green icon if everything's checked out. So, for example, if someone is hiding the barcode, but just still swipe, do the swipe action of a bottle, then the camera will see the bottle, but the barcode scanner wouldn't catch it. And so, by that determination, something might be wrong. So, this is just one example of combining EdgeX data stream processing and event processing to be able to achieve a theft detection use case. We are able to work with all these partners, supplying the software, supplying the devices, HP working with Intel, IoTeX, and many other partners to be able to put together this demonstration very quickly within a matter of two months. So, further on, we can extend the same model, give an example of if we wanted to do a quick service restaurant that is the fast food restaurant to do an inventory checkout, inventory tracking. We can make use of RFID, for example, to watch for items being delivered, move into storage, using kitchen, and then use in the dining room. These could be the hamburger patties, the drink cups, the straws, ketchup packets, etc. There are just many, many items to be tracked within a quick service restaurant. So, these can all be based on the same EdgeX core, except of course the data stream is coming in as via RFID. In this example that RFID events can also be fed into EdgeX via the MQTT interface, and you can still have, create a similar reconciler to determine the results, and you can present the results in a UI dashboard or you can further compare the results to some other server side database where you have to keep track of things in order to determine when you need to reorder certain items in your inventory. These again, the core remains the same, and you can replace a few software components and cover a totally different use case. So that's the purpose of EdgeX, right? It is a highly flexible open software framework that allows interoperability between devices and applications at the IoT Edge. EdgeX is hosted by the Linux Foundation and this project has been around for three years. Within this project, we also have a vertical solutions working group. For the past two to three months, we have various industry players come in to present how they are using EdgeX, how they are making use of EdgeX to build their product, what do they like about EdgeX, and maybe even in the future, what additional features they are seeking in EdgeX. So, feel free to check out that working group to look at the previous session recordings. And in HP, about a month ago, we announced this Engage Edge product, which is a Edge gateway that incorporates EdgeX. And the purpose is to have this Edge gateway element to allow the ISV partners to accelerate a Edge deployment in retail, hospitality, and many other use cases. So feel free to look into that. So thank you very much and happy to answer any questions. Thank you.