Building Video AI Applications

 

About this Course

AI-based video understanding can unlock insights, whether it’s recognizing a cat in your backyard or optimizing customers’ shopping experiences. The NVIDIA Jetson Nano Developer Kit is an easy-to-use, powerful computer that lets you run multiple neural networks in parallel. This makes it a great platform for an introduction to intelligent video analytics (IVA) applications using the NVIDIA DeepStream SDK. In this course, you'll use JupyterLab notebooks and Python application samples on your Jetson Nano to build new projects that extract meaningful insights from video streams through deep learning video analytics. The techniques you learn from this course can then be applied to your own projects in the future on the Nano or other Jetson platforms at the Edge.

Learning Objectives

You'll learn how to:

Set up your Jetson Nano

Build end-to-end DeepStream pipelines to convert raw video input into insightful annotated video output

Build alternate input and output sources into your pipeline

Configure multiple video streams simultaneously

Configure alternate inference engines such as YOLO

Upon completion, you'll be able to build DeepStream applications that annotate video streams from various and multiple sources to identify and classify objects, count objects in a crowded scene, and output the result as a live stream or file.

Topics Covered

Tools, libraries, frameworks used in this course include DeepStream, TensorRT, Jetson Nano, and Python

Course Outline

1. Setting up your Jetson Nano

Step-by-step guide to set up your hardware and software for the course projects

Note: This course supports the NVIDIA Jetson Nano Developer Kit but does not support the NVIDIA Jetson Orin Nano Developer Kit

• Introduction and Setup

Video walk-through and instructions for setting up JetPack and what items you need to get started

• Camera Setup

How to connect your camera to the Jetson Nano Developer Kit

• Headless Device Mode

Video walk-through and instructions for running the Docker container for the course using headless device mode (remotely from your computer).

• JupyterLab

A brief introduction to the JupyterLab interface and notebooks

• Media Player

How to set up video streaming on your computer

2. Introduction to DeepStream SDK

Overview of key DeepStream SDK features and important reference links for deeper exploration

• What is the DeepStream SDK?

An overview of DeepStream applications and the DeepStream SDK

• GStreamer Plugins

Introduction to the GStreamer framework and plugins

• TensorRT

Introduction to TensorRT

• Video to Analytics With the DeepStream SDK

Outline of the DeepStream metadata structure

3. Exploring DeepStream SDK

Course notebook and environment details for your Jetson Nano hands-on learning experience

• Build DeepStream Applications

Instructions for opening the first notebook in JupyterLab on Jetson Nano

• Exercises

A summary of the lesson notebooks included in the Jetson Nano MicroSD card image.

• Directory Structure

A summary of the DeepStream SDK directory structure

Free Courses : Building Video AI Applications