Skip to content

Docker on Windows

One way of installing Docker on Windows is using Docker Desktop. However, if you are not afraid of running a few commands in the terminal and having a terminal open while working with Docker, then you can follow these instructions of getting Docker working under WSL2.


NB: The following WSL commands are to be executed in a Windows command-prompt.

  • Make sure that WSL is up to date: wsl --update
  • You can switch to version 2 of WSL by default with: wsl --set-default-version 2
  • To check what version any current images are running under, use: wsl --list -v
  • For converting an existing image called Ubuntu-22.04 to WSL 2 use: wsl --set-version Ubuntu-22.04 2


For more information on how to configure WSL2, please see here.

Installing Docker#

  • Get Ubuntu 20.04.x or Ubuntu 22.04.x from the Windows store
  • Configure the default user and password when asked for during the installation
  • Get your system ready
sudo apt update && sudo apt upgrade
sudo apt install --no-install-recommends apt-transport-https ca-certificates curl gnupg2
  • Change iptables to legacy:
sudo update-alternatives --config iptables
  • Install Docker
. /etc/os-release
curl -fsSL${ID}/gpg | sudo tee /etc/apt/trusted.gpg.d/docker.asc
echo "deb [arch=amd64]${ID} ${VERSION_CODENAME} stable" | sudo tee /etc/apt/sources.list.d/docker.list
sudo apt update
sudo apt install docker-ce docker-ce-cli
sudo usermod -aG docker $USER
  • Close the WSL2 window and open a new one for the changes to take effect
  • Test your Docker installation by running:
sudo dockerd
  • After a lot of output of Docker starting up, you should see the following:
API listen on /var/run/docker.sock

(The above instructions were taken from this post by Jonathan Bowman)

Installing NVIDIA Docker#

Since our CUDA dependencies are packaged within the Docker images, we only need to worry about installing NVIDIA Docker to get access to the GPU:

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L | sudo apt-key add -
curl -s -L$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit

Due to installing a new runtime, we need to restart our dockerd daemon, of course.

(The above instructions were taken from this post by Frank Chung)

Testing the GPU#

The following commands test the inference of a Yolov7 model on the GPU:

  • Create a test directory:
mkdir gpu-test
cd gpu-test
  • Download the pre-trained model and a test image:
wget ""
  • Perform inference:
docker run --rm -u $(id -u):$(id -g) \
  -v `pwd`:/workspace \
  -w /workspace \
  --gpus=all \
  -t waikatodatamining/pytorch-yolov7:2022-10-08_cuda11.1 \
  yolov7_detect \
  --weights ./ \
  --source ./'2021_Toyota_GR_Yaris_Circuit_4WD_1.6_(1).jpg' \
  --no-trace \
  --conf-thres 0.8 \
  --device 0
  • In directory gpu-test/runs/detect/exp you will find the image with the detected objects overlaid.