L4T based Docker container for NVIDIA Jetson devices.
Including TensorFlow (python) and GStreamer Support for OpenCV (required when running gstreamer commands through the opencv api, e.g. with the cv2 python package).
To run a docker container with GPU support on a Jetson, run your docker container with --runtime=nvidia or change the default runtime. If you need to access NVIDIA libraries (such as e.g. CUDNN) during build of your container you will need to change the default runtime, as the --runtime flag is not available for docker build.
docker build . -t jetson
docker run -v /tmp/argus_socket:/tmp/argus_socket \
-it jetson
Mounting /tmp/argus_socket will allow to access any cameras connected to the Jetson.
Tested on NVIDIA Jetson Xavier NX.