In this recipe, we added the libraries and then we cloned the Jetson inference repository. We then ran a series of make and linker tools to get the install working correctly. During this process, we downloaded a large set of pretrained models. We then started writing our code. Because the Jetson is limited in terms of its capabilities and memory, installing a full-featured IDE can be wasteful. One workaround for this is to use an IDE that supports SSH, such as Visual Studio Code, and remoting into the box via the IDE. You can then work with the device without tying up resources on the Jetson Nano.
To build out this project, first, we import the Jetson inference and utils libraries. In the previous recipes, we did a lot of the low-level work ourselves as far as using OpenCV to get the camera and then used other libraries to manipulate the images and draw bounding boxes. With Jetson's libraries, the vast majority of that code is handled for you. After we imported...