See a demonstration of facial detection and recognition using the Intel® Computer Vision SDK and the OpenCV library. Learn how to use these tools for security, smart retail, smart cities, and robotics applications.
Hi, I'm Martin Kronberg, and this is the IOT Developers' Show where we look at IOT technology, shared learning opportunities, and showcase cool demos and the creators behind them. In this episode, we're going to be taking a look at an awesome facial recognition and detection demo with Tudor Panu, a software engineer here at Intel. Let's get started.
Tudor, thank you so much for joining us. What do you do here at Intel?
Hi, I've been a software engineer with Intel for a little over three years now, and I focus mostly on tools for IOT developers. Before Intel, I got a PhD in computer science and engineering from SMU in Dallas, Texas.
All right, cool. Well, do you want to tell us a bit more about the demo that you have for us?
This application is called face access control and can be used for facial detection and recognition. There are two main components, a video streaming service, which includes the computer vision elements, and a user based interface for registration and analytics. In this set up, I used Intel's Computer Vision SDK and a sixth generation Core i7 Intel NUC with a web camera.
Awesome. Well, I'd love to see how it all works. Let's check it out.
Let's take a look.
I'm going to use my own face to show you how it works. First, you have to add your information to the database. All you have to do is click the new profile button when you're ready to save your picture. Once the camera sees a face, it's already tracking you. Thus, you can register a new person in real time with a single click. Here, I am entering the information in the database to get through registration. Since not all fields are mandatory, I will just add my name and clearance level.
OK. So you can basically assign different access levels to people depending on their roles.
Exactly. The clearance level really asks you to access control side of the story and can be used to trigger events such as opening a door or sounding an alarm.
Let's see how it works.
You can see my profile picture pop up there on the left with my name.
OK, it looks like it can easily recognize you.
Yeah, it's pretty accurate. Let me show you the analytics portion of the application. Every time a face gets detected, a new data point is added to the chart. You can also look at exact counts and distributions on this screen. Timestamps are also available on the side panel.
All right. Well, I really like this interface that you've created. How do you think the developers can use an application like this?
So, the application will obviously be used in digital surveillance and security. However, there are some other verticals that can be touched with it such as smart retail, smart cities and buildings, and even robotics. For instance, think of membership based services like a gym. You might want to send push notifications to your clients whenever they don't show up for a while, or if you have a large retail store and you have a list of banned people, you might want to alert security whenever somebody from that list shows up.
Sure. Maybe like known shoplifters that have been caught in the past.
So, you mentioned that this application is built using the CVSDK. Is that right?
Correct. We do use the open CV version that comes with Intel's Computer Vision SDK, and that library also includes the facial detection and recognition algorithms. These are faster than traditional methods, and we can do detection or recognition almost on every frame with low light SD.
So, what kind of system would I need if I wanted to run this demo?
Any old bluetooth 1604 system that supports the Computer Vision SDK will run it, and you can also scale from Atom to Core to Xeon based on your performance requirements.
Wow. Well, this is a really cool Computer Vision demo. If you guys are more interested in learning about the specifics of the demo or how to apply it into your own project, we're going to provide all of the links. Tudor, thank you so much for being here and showing this demo off.
Thanks for having me, Martin.
Before signing off, I want to tell you guys about some IOT news. Recently, we launched the Intel E workshop. Now, this is a virtual lab that walks you through the set up of an Intel LG gateway with a sensor kit to be able to read sensor values and then publish that data to the cloud. Also, recently we launched the Up Squared Grove IOT Developer kit. Go to the Intel Developer Zone for IOT to get resources and tutorials to be able to help you reduce development time for your IOT applications.
Thanks for we're watching this episode of the IOT Developer Show. Don't forget to like this video and subscribe to the Intel software YouTube channel. Remember to check out the resources and the links provided as well. Thanks, guys.