You then had to run TensorFlow Lite models on smartphones. How about on lightweight embedded devices like this. This is a Raspberry Pi Zero which is very lightweight, low-power, pretty much full flesh computer. If you can get a neural network to run on this, you can start to build intelligent devices ranging from smart doorbells, maybe have rudimentary speech recognition in your desk lamp. Yes. So the sky's the limit. Absolutely, and even the little self-driving car that we showed earlier on in the course. So one of the things that we really want to help students learn is that part, is getting neural network, getting models to run on devices like these. That's a bigger Raspberry Pi than this one. So this is a standard Raspberry Pi rather than the Raspberry Pi Zero. I like the little one. I always say this one's like a credit card and that one's like a stick of gum. But one of the things that's really impressive and I love about these is the shear scenario of things that you can do with them, as Andrew mentioned, like maybe smart doorbells and I built that little self-driving car. But in the course, what you will learn to do is to put a couple of models on here that use the TensorFlow Lite Interpreter. So the first model that you'll put on is just the very basic image classification. So you can take a static JPEG and you'll run it through a classifier, and then you'll get results back about what the JPEG is, what the contents of that JPEG are. But then after that, even more exciting is the Raspberry Pi has a little interface here for a camera. So you can put a camera module onto that, and then you'll do live object detection of what that cameras actually seeing. So you'll write the code for that and you'll see how it all works and how it all hangs together. Hopefully, that will give you the platform on which you can start building really amazing things with systems like Raspberry Pi. Again, if they have a Raspberry Pi or a Pi Zero, they will run this on their own devices. But if they don't have one of these cool devices, then they'll simply do [inaudible] using a simulator. Yes. So the simulator that you will use in this case will be in Colab. You can run the TensorFlow Lite Interpreter. So all of the code that we'll be showing in this that you could then just run through the TensorFlow Lite Interpreter in Colab. So yeah, it should be exciting to work with. Yeah. I sometimes find it one of the most fun things to do on the weekends. I hack around on small devices like these. So I think you enjoy this week's materials.