/Jetson_Ubuntu20_Images

Hardware accelerated Docker images based on Ubuntu 20.04 for Jetson family

Primary LanguageShell

Deprecation notice

This repository will no longer be updated, since Nvidia started to provide its own Docker images based on Ubuntu 20 as of JETPACK SDK 5.0.1 (Link to images).

Therefore, the Dockerfiles in this repository should only be used as reference if CUDA 10.2 is a requirement.

Hardware accelerated Docker images based on Ubuntu 20.04 for Jetson family

Nvidia Jetson images are based on Ubuntu 18.04. However, many applications and projects utilizes libraries specific to Ubuntu 20.04. Therefore, this repository provides docker images based on ubuntu:focal, that are able to take full advantage of the Jetson hardware (Nano, Xavier NX, Xavier AGX and Xavier TX2). All images come with full CUDA support (passthrough from the host) including TensorRT (including python bindings) and VisionWorks.

Furthermore the script folder includes system installable run-scripts to quickly iterate the build process and set the relevant docker flags at runtime.

The following images can be directly pulled from DockerHub without needing to build the containers yourself:

L4T Version Dockerhub image
l4t-ubuntu20-base R32.7.1 timongentzsch/l4t-ubuntu20-base:latest
l4t-ubuntu20-opencv R32.7.1 timongentzsch/l4t-ubuntu20-opencv:latest
l4t-ubuntu20-pytorch R32.7.1 timongentzsch/l4t-ubuntu20-pytorch:latest
l4t-ubuntu20-ros2-base R32.7.1 timongentzsch/l4t-ubuntu20-ros2-base:latest
l4t-ubuntu20-ros2-desktop R32.7.1 timongentzsch/l4t-ubuntu20-ros2-desktop:latest
l4t-ubuntu20-zedsdk R32.7.1 timongentzsch/l4t-ubuntu20-zedsdk:latest
l4t-ubuntu20-crosscompile R32.7.1 timongentzsch/l4t-ubuntu20-crosscompile:latest

note: make sure to run the container on the intended L4T host system. Running on older JetPack releases (e.g. r32.6.1) can cause driver issues, since L4T drivers are passed into the container.

note: pytorch image also inckudes the python3.8 tensorrt bindings

To download and run one of these images, you can use the included run script from the repo:

$ scripts/docker_run timongentzsch/l4t-ubuntu20-base:latest

For other configurations, below are the instructions to build and test the containers using the included Dockerfiles.

Docker Default Runtime

To enable access to the CUDA compiler (nvcc) during docker build operations, add "default-runtime": "nvidia" to your /etc/docker/daemon.json configuration file before attempting to build the containers:

{

"runtimes": {

  "nvidia": {

    "path": "nvidia-container-runtime",

    "runtimeArgs": []

  }

 },

"default-runtime": "nvidia"

}

You will then want to restart the Docker service or reboot your system before proceeding.

Build and test the images

To rebuild the containers from a Jetson device, first clone this repo via Git LFS:

$ git clone https://github.com/timongentzsch/Jetson_Ubuntu20_Images.git

$ cd Jetson_Ubuntu20_Images

Work with the provided scripts

You may want to install the provided scripts to build, run and restart containers with the right set of docker flags:

$ sudo scripts/setup_launchscripts.sh

This will enable you to quickly iterate your build process and application. Sudo is needed to create the dummy_root_SSH_config and enable SSH access for root users

After that you can use following commands globally:

dbuild, drun, dstart

It ensures that the docker environment feels as native as possible by enabling the following features by default:

  • USB hot plug

  • sound

  • network

  • bluetooth

  • GPU/cuda

  • X11

note: refer to --help for the syntax