Back | Next | Contents
System Setup
note: if your Jetson Nano or Xavier NX has already been setup with the SD card image (which includes JetPack), or your Jetson has already been setup with JetPack, you can skip this step and continue to
Running the Docker ContainerorBuilding the Project
NVIDIA JetPack is a comprehensive SDK for Jetson for both developing and deploying AI and computer vision applications. JetPack simplifies installation of the OS and drivers and contains the following components:
- L4T Kernel / BSP
- CUDA Toolkit
- cuDNN
- TensorRT
- OpenCV
- VisionWorks
- Multimedia API's
Before attempting to use the Docker container or build the repo, make sure that your Jetson has been setup with the latest version of JetPack.
The recommended install method for the Jetson Nano Developer Kit and Jetson Xavier NX Developer Kit is to use the SD card images.
It comes pre-populated with the JetPack components already installed and can be flashed from a Windows, Mac, or Linux PC. If you haven't already, follow the Getting Started guide for your respective Jetson to flash the SD card image and setup your device:
- Getting Started with Jetson Nano Developer Kit
- Getting Started with Jetson Nano 2GB Developer Kit
- Jetson Xavier NX User Guide
Other Jetson's should be flashed by downloading the NVIDIA SDK Manager to a host PC running Ubuntu 16.04 x86_64 or Ubuntu 18.04 x86_64. Connect the Micro-USB or USB-C port to your host PC and enter the device into Recovery Mode.
For more details, please refer to the NVIDIA SDK Manager Documentation.
There are two ways to use the jetson-inference project:
- Run the pre-built Docker Container
- Build the Project from Source
Using the container is recommended initially to get up & running as fast as possible (and the container already includes PyTorch installed), however if you are more comfortable with native development then compiling the project yourself is not complicated either.
Next | Building the Project from Source
Back | Overview
© 2016-2019 NVIDIA | Table of Contents