Machine Learning Containers for Jetson and JetPack

Hosted on NVIDIA GPU Cloud (NGC) are the following Docker container images for machine learning on Jetson:

The following ROS containers are also provided, which can be pulled from DockerHub or built for JetPack 4.4 or newer:

ROS Melodic (ros:melodic-ros-base-l4t-r34.1.1)
ROS Noetic (ros:noetic-ros-base-l4t-r34.1.1)
ROS2 Eloquent (ros:eloquent-ros-base-l4t-r34.1.1)
ROS2 Foxy (ros:foxy-ros-base-l4t-r34.1.1)
ROS2 Galactic (ros:galactic-ros-base-l4t-r34.1.1)
ROS2 Humble (ros:humble-ros-base-l4t-r34.1.1)

Pre-built Container Images

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

	L4T Version	Container Tag
`l4t-ml`	R34.1.0	`nvcr.io/nvidia/l4t-ml:r34.1.0-py3`
	R32.7.1	`nvcr.io/nvidia/l4t-ml:r32.7.1-py3`
	R32.6.1	`nvcr.io/nvidia/l4t-ml:r32.6.1-py3`
	R32.5.0*	`nvcr.io/nvidia/l4t-ml:r32.5.0-py3`
	R32.4.4	`nvcr.io/nvidia/l4t-ml:r32.4.4-py3`
	R32.4.3	`nvcr.io/nvidia/l4t-ml:r32.4.3-py3`
`l4t-pytorch`	R34.1.0	`nvcr.io/nvidia/l4t-pytorch:r34.1.0-pth1.12-py3`
	R32.7.1	`nvcr.io/nvidia/l4t-pytorch:r32.7.1-pth1.10-py3`
	R32.7.1	`nvcr.io/nvidia/l4t-pytorch:r32.7.1-pth1.9-py3`
	R32.6.1	`nvcr.io/nvidia/l4t-pytorch:r32.6.1-pth1.9-py3`
	R32.6.1	`nvcr.io/nvidia/l4t-pytorch:r32.6.1-pth1.8-py3`
	R32.5.0*	`nvcr.io/nvidia/l4t-pytorch:r32.5.0-pth1.7-py3`
	R32.5.0*	`nvcr.io/nvidia/l4t-pytorch:r32.5.0-pth1.6-py3`
	R32.4.4	`nvcr.io/nvidia/l4t-pytorch:r32.4.4-pth1.6-py3`
	R32.4.3	`nvcr.io/nvidia/l4t-pytorch:r32.4.3-pth1.6-py3`
`l4t-tensorflow`	R34.1.0	`nvcr.io/nvidia/l4t-tensorflow:r34.1.0-tf1.15-py3`
	R34.1.0	`nvcr.io/nvidia/l4t-tensorflow:r34.1.0-tf2.8-py3`
	R32.7.1	`nvcr.io/nvidia/l4t-tensorflow:r32.7.1-tf1.15-py3`
	R32.7.1	`nvcr.io/nvidia/l4t-tensorflow:r32.7.1-tf2.7-py3`
	R32.6.1	`nvcr.io/nvidia/l4t-tensorflow:r32.6.1-tf1.15-py3`
	R32.6.1	`nvcr.io/nvidia/l4t-tensorflow:r32.6.1-tf2.5-py3`
	R32.5.0*	`nvcr.io/nvidia/l4t-tensorflow:r32.5.0-tf1.15-py3`
	R32.5.0*	`nvcr.io/nvidia/l4t-tensorflow:r32.5.0-tf2.3-py3`
	R32.4.4	`nvcr.io/nvidia/l4t-tensorflow:r32.4.4-tf1.15-py3`
	R32.4.4	`nvcr.io/nvidia/l4t-tensorflow:r32.4.4-tf2.3-py3`
	R32.4.3	`nvcr.io/nvidia/l4t-tensorflow:r32.4.3-tf1.15-py3`
	R32.4.3	`nvcr.io/nvidia/l4t-tensorflow:r32.4.3-tf2.2-py3`
`ROS Melodic`	R32.7.1	`dustynv/ros:melodic-ros-base-l4t-r32.7.1`
	R32.6.1	`dustynv/ros:melodic-ros-base-l4t-r32.6.1`
	R32.5.0*	`dustynv/ros:melodic-ros-base-l4t-r32.5.0`
	R32.4.4	`dustynv/ros:melodic-ros-base-l4t-r32.4.4`
`ROS Noetic`	R34.1.1	`dustynv/ros:noetic-ros-base-l4t-r34.1.1`
	R34.1.0	`dustynv/ros:noetic-ros-base-l4t-r34.1.0`
	R32.7.1	`dustynv/ros:noetic-ros-base-l4t-r32.7.1`
	R32.6.1	`dustynv/ros:noetic-ros-base-l4t-r32.6.1`
	R32.5.0*	`dustynv/ros:noetic-ros-base-l4t-r32.5.0`
	R32.4.4	`dustynv/ros:noetic-ros-base-l4t-r32.4.4`
`ROS2 Eloquent`	R32.7.1	`dustynv/ros:eloquent-ros-base-l4t-r32.7.1`
	R32.6.1	`dustynv/ros:eloquent-ros-base-l4t-r32.6.1`
	R32.5.0*	`dustynv/ros:eloquent-ros-base-l4t-r32.5.0`
	R32.4.4	`dustynv/ros:eloquent-ros-base-l4t-r32.4.4`
`ROS2 Foxy`	R34.1.1	`dustynv/ros:foxy-ros-base-l4t-r34.1.1`
	R34.1.0	`dustynv/ros:foxy-ros-base-l4t-r34.1.0`
	R32.7.1	`dustynv/ros:foxy-ros-base-l4t-r32.7.1`
	R32.6.1	`dustynv/ros:foxy-ros-base-l4t-r32.6.1`
	R32.5.0*	`dustynv/ros:foxy-ros-base-l4t-r32.5.0`
	R32.4.4	`dustynv/ros:foxy-ros-base-l4t-r32.4.4`
`ROS2 Galactic`	R34.1.1	`dustynv/ros:galactic-ros-base-l4t-r34.1.1`
	R34.1.0	`dustynv/ros:galactic-ros-base-l4t-r34.1.0`
	R32.7.1	`dustynv/ros:galactic-ros-base-l4t-r32.7.1`
	R32.6.1	`dustynv/ros:galactic-ros-base-l4t-r32.6.1`
	R32.5.0*	`dustynv/ros:galactic-ros-base-l4t-r32.5.0`
	R32.4.4	`dustynv/ros:galactic-ros-base-l4t-r32.4.4`
`ROS2 Humble`	R34.1.1	`dustynv/ros:humble-ros-base-l4t-r34.1.1`
	R34.1.0	`dustynv/ros:humble-ros-base-l4t-r34.1.0`

note: the L4T R32.5.0 containers can be run on both JetPack 4.5 (L4T R32.5.0) and JetPack 4.5.1 (L4T R32.5.1)

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

# L4T version in the container tag should match your L4T version
$ scripts/docker_run.sh -c nvcr.io/nvidia/l4t-pytorch:r32.5.0-pth1.7-py3

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.

Building the Containers

To rebuild the containers from a Jetson device running JetPack 4.4 or newer, first clone this repo:

$ git clone https://github.com/dusty-nv/jetson-containers
$ cd jetson-containers

Before proceeding, make sure you have set your Docker Default Runtime to nvidia as shown above.

ML Containers

To build the ML containers (l4t-pytorch, l4t-tensorflow, l4t-ml), use scripts/docker_build_ml.sh - along with an optional argument of which container(s) to build:

$ ./scripts/docker_build_ml.sh all        # build all: l4t-pytorch, l4t-tensorflow, and l4t-ml
$ ./scripts/docker_build_ml.sh pytorch    # build only l4t-pytorch
$ ./scripts/docker_build_ml.sh tensorflow # build only l4t-tensorflow

You have to build l4t-pytorch and l4t-tensorflow to build l4t-ml, because it uses those base containers in the multi-stage build.

Note that the TensorFlow and PyTorch pip wheel installers for aarch64 are automatically downloaded in the Dockerfiles from the Jetson Zoo.

ROS Containers

To build the ROS containers, use scripts/docker_build_ros.sh with the --distro option to specify the name of the ROS distro to build:

$ ./scripts/docker_build_ros.sh --distro all       # build all of the below (default)
$ ./scripts/docker_build_ros.sh --distro melodic   # build only melodic
$ ./scripts/docker_build_ros.sh --distro noetic    # build only noetic
$ ./scripts/docker_build_ros.sh --distro eloquent  # build only eloquent
$ ./scripts/docker_build_ros.sh --distro foxy      # build only foxy
$ ./scripts/docker_build_ros.sh --distro galactic  # build only galactic

You can also specify --with-pytorch and --with-slam to build variants with support for PyTorch and GPU-accelerated SLAM nodes (including ORBSLAM2 and RTABMAP). Note that Noetic, Foxy, and Galactic are built from source for Ubuntu 18.04, while Melodic and Eloquent are installed from Debian packages into the containers.

Testing the Containers

To run a series of automated tests on the packages installed in the containers, run the following from your jetson-containers directory:

$ ./scripts/docker_test_ml.sh all        # test all: l4t-pytorch, l4t-tensorflow, and l4t-ml
$ ./scripts/docker_test_ml.sh pytorch    # test only l4t-pytorch
$ ./scripts/docker_test_ml.sh tensorflow # test only l4t-tensorflow

To test ROS:

$ ./scripts/docker_test_ros.sh all       # test if the build of ROS all was successful: 'melodic', 'noetic', 'eloquent', 'foxy'
$ ./scripts/docker_test_ros.sh melodic   # test if the build of 'ROS melodic' was successful
$ ./scripts/docker_test_ros.sh noetic    # test if the build of 'ROS noetic' was successful
$ ./scripts/docker_test_ros.sh eloquent  # test if the build of 'ROS eloquent' was successful
$ ./scripts/docker_test_ros.sh foxy      # test if the build of 'ROS foxy' was successful

Name		Name	Last commit message	Last commit date
Latest commit History 199 Commits
packages		packages
scripts		scripts
test		test
.dockerignore		.dockerignore
.gitignore		.gitignore
Dockerfile.jetpack		Dockerfile.jetpack
Dockerfile.ml		Dockerfile.ml
Dockerfile.opencv		Dockerfile.opencv
Dockerfile.pytorch		Dockerfile.pytorch
Dockerfile.ros.eloquent		Dockerfile.ros.eloquent
Dockerfile.ros.foxy		Dockerfile.ros.foxy
Dockerfile.ros.galactic		Dockerfile.ros.galactic
Dockerfile.ros.humble		Dockerfile.ros.humble
Dockerfile.ros.melodic		Dockerfile.ros.melodic
Dockerfile.ros.noetic		Dockerfile.ros.noetic
Dockerfile.ros.slam		Dockerfile.ros.slam
Dockerfile.tensorflow		Dockerfile.tensorflow
LICENSE.md		LICENSE.md
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Machine Learning Containers for Jetson and JetPack

Pre-built Container Images

Docker Default Runtime

Building the Containers

ML Containers

ROS Containers

Testing the Containers

About

Releases

Packages

Languages

License

prl-mushr/jetson-containers

Folders and files

Latest commit

History

Repository files navigation

Machine Learning Containers for Jetson and JetPack

Pre-built Container Images

Docker Default Runtime

Building the Containers

ML Containers

ROS Containers

Testing the Containers

About

Resources

License

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages