Deep Learning APIs Installation - jad-mansour/jetson-inference GitHub Wiki

Before going through this section, make sure you have all dependencies, such as CUDA, installed using SDKManager.

The most straightforward method to install all required packages is by using a Docker image inside the Jetson. Follow the installation steps below:

  1. Jetpack OS comes with Docker installed. However, Docker is not run by default as a super user. To do that, run the command:
sudo usermod -aG docker $USER
  1. Restart your session and configure the NVIDIA runtime container as shown below:
sudo nvidia-ctk runtime configure --runtime=docker

  1. Now that we have docker ready, it is time to run one of the docker images from the NVIDIA L4T-ML catalog. Make sure to choose a docker image that is compatible with your JetPack version.

  2. You can add the -v argument to connect a host directory on the Jetson to the Docker image directory.

Now you have all necessary packages installed to run your inferences on the Jetson!