Jetson Docker OpenCV CUDA

---

• Added support Jetpack 4.6
• Build tested on Jetson Nano & Jetson Xavier

---

• Set nvidia as default Docker runtime,

sudo nano /etc/docker/daemon.json

• Change daemon.json to look like this,

{
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
         } 
    },
    "default-runtime": "nvidia" 
}

• Restart Docker Service,

sudo systemctl restart docker

• Clone repo :

cd ~
mkdir -p Github
cd Github
git clone https://github.com/Muhammad-Yunus/Jetson-Docker-OpenCV-CUDA.git

• Build image,

cd Jetson-Docker-OpenCV-CUDA/docker
sudo chmod +x build.sh
./build.sh

• Run Image using NVIDIA Container Runtime,

sudo docker run --rm --net=host --runtime nvidia  opencv-cuda:latest

• Or, Pull image from Docker Hub :

docker pull yunusdev/jetson-opencv-cuda

• Run Image using NVIDIA Container Runtime,

sudo docker run --rm --net=host --runtime nvidia  yunusdev/jetson-opencv-cuda:latest

Building on a different base container

By default, the build script builds on top of a container matching the version of Jetpack running on the host. To build with a specific base image instead, use:

./docker/build.sh --image <your base image>

For example, to build on the tensorflow-l4t container use a command like:

./docker/build.sh --image nvcr.io/nvidia/l4t-tensorflow:r32.7.1-tf2.7-py3