 All right. Hello and welcome to our presentation, ladies and gentlemen. Our project is your door, and we are going to be presenting this to you guys. It's a state-of-the-art facial recognition technology on doors around campuses. All right, my name is Peter Schwartz, and this is my partner and colleague, Jacob Johnson. Okay, so currently you're walking around campus and somebody loses their key card for their dorm. Somebody can walk up, find that key card laying on the ground, grab it, and use it on a door. This would allow unsecure access to buildings for that other person. The other problem that also exists is tailgating. So if somebody was to see somebody accessing a door, and then they walked up behind them and also gained entry into that into that door frame, then they would have unauthorized access as well. So another problem that both of these create is theft of college property. All right. So our solution? Your door. Your door offers state-of-the-art facial recognition technology, and it's safe, secure, and effortless. It also houses a standalone server, which will keep the integrity of the data intact and with the customer, which is very, very important. All right, so we would initially target colleges and universities. We have pretty good knowledge of those systems as is. I currently work for GTS, Gustavus Technology Services, and my partner, Jacob, he works for Campus Safety right now. We then would target the small businesses. We would look at targeting the small businesses, mainly to track analytics, and also to grab some marketing data for them. So at this point you're probably wondering what goes into your door. So there are two main areas of our product. The first are the door components and the second are the main frame components. The door components consist of a camera to record the faces, a Raspberry Pi, which is a little computer chip, that will relay the faces to our facial recognition software and receive an input from that, and then finally an electronic lock, which will then relay back to the Raspberry Pi and will either unlock or not, depending on if the face was recognized. We then have our main frame components. These are things that will only need one of per customer. And these are the central server, which houses our facial recognition software as well as our custom tailored software that we would design. So we're going to take an example here. We're going to take Nobel Hall of Science, which is currently one of our most at-risk buildings on campus, due to the liability of the expensive equipment inside. Now we're going to take an example of the front, one of the front doors, and how this would look. So, as you can see, we would set up a camera on the right in the center. You can see the two electronic door locks and the Raspberry Pi that is housed either in the ceiling or in the door frame nearby. This would then have two modes, an active mode and a passive mode. The active mode would work at night when the doors are supposed to be locked and would only allow certain users access to the building such as campus safety. We would then also have a passive mode during the day, which would allow anybody into the building, as the building is right now, but it would keep track of people who are not registered with the college, so non-students and non-faculty. We would then also set up in campus safety, or similar location, our central server with our facial recognition software, as well as our custom tailored application, which would allow campus safety to see the individuals who have been flagged throughout the day. At this point, you might be wondering, who are our competitors? So, our two main competitors are facial recognition software companies, such as Face First, and key card reader companies. We've already shown that we have some clear advantages over key card reader companies in terms of security, as well as customizability, but we also have some clear advantages over current facial recognition software companies. The first is commercial availability, the fact that the majority of companies right now are offering this only to large corporations or government contracts, whereas we can provide this to smaller businesses as well as colleges and universities. Second is that we offer localized data, so the big thing right now is cloud storage. We would use a central server instead, which keeps the data with the customers, as opposed to a company such as Microsoft, which would then have you input the photos to their cloud storage, and they would effectively own your data. Finally, we have the customizability of our product, which is the active and passive mode, as well as many other things that we could build on top of the facial recognition software. Yeah, so our management team currently consists of myself and Jacob, and we are both computer science managers and we both have experience in facial recognition technology. I've worked at Mayo Clinic doing some facial recognition stuff on some apps of theirs, and Jacob has done some research at Oxford University from his study abroad. So our first target would probably be a marketing expert. We need somebody with a know-how to help us with the marketing side of things. We have the software side of things, but we also need the marketing. So now let's take a little bit at our marketing and sales. We once again would initially target colleges and universities, that's kind of our buy-in. Now one of the things we can also provide is a potential reduction in terms of liability insurance. So let's take Gustavus for an example. Gustavus is currently paying approximately $180,000 a year for liability insurance to cover their 1,300,000 square feet. Now say we could offer them a 3 cent discount per square foot, right there we are saving them $39,000 a year, which practically pays for our system. On top of that, we also buy all of the parts and would offer installation of all of the physical components, as well as build the custom tailored software ourselves, which would in essence mean that our customer does not have to have the technical experience necessary to implement this system. We would do it all for them. So for our financial projections, there's initially a $50,000 opportunity cost of building the facial recognition software. But then after that, we can see that by year four, when we have estimated that we would be in approximately 30 institutions, we are at almost a quarter of a million dollars in profit. Alright, so we would seek some funding to initially build the software for our company. The core components of the software would have to have some funding. We would also need some financing for the marketing side of things. And finally, most of those costs will be covered by the customer. So the customer will take care of installation and the core components. So people across the world have been looking for a new way of secure, safe, and effortless access to buildings. And we think that your door is the solution that campuses across the world need. I guess I've been involved with some factor authentication security systems in the past. The obvious question is, if somebody came up to the door with a picture, has that been tested to prove that... Yeah, so many facial recognition algorithms currently have undergone a lot of different testing. And I had the experience to work with some researchers that are kind of leading edge in this field. And right now, almost all of them can detect if it's just somebody holding up a picture. There have been a couple of issues with identical twins, but right now that's the biggest obstacle that they're trying to overcome in facial recognition. Yeah, one simple thing for that is if you check for blinking. Blinking is one of the easiest ways to automatically get rid of a fake, like held up picture. Like if somebody's pulling up a piece of paper. Okay, but then you get a little bit more elaborate. Say you have an ass that blinks or things like that. Sure. So right now the majority of facial recognition algorithms have 99.999% accuracy in terms of being able to recognize across many test cases. It all depends on how you build it. And you end up training on a lot of images already. But it is something that is very highly efficient right now and very, very secure. Are you using off the shelf hardware and then are you going to build the algorithm? Yeah, so we would build the algorithm from scratch. And so that would be the biggest cost to us initially is actually coming up with designing that and then testing it, making sure that it works in all of our use cases. We would use off the shelf hardware as well. We would try and partner with specific hardware vendors so that we can ensure like the same consistency across all of our institutions. Let's hear it for them.