Use your webcam to unlock hand gesture recognition to control Neovim!

Highly experimental. Is it practical? Maybe not, but some use cases are not bad. Should you try it? Why not, it's fun!

Highly customisable. Just change the lua scripts to your own custom commands!
The demo uses GitHub Copilot (AI code suggestion), Nvim-Tree (file explorer), and nvim-treesitter-textobjects (peep function definition).

You need an Nvidia GPU on a Linux computer. I tested with Ubuntu 22.04, GTX 1080 Ti and RTX 3060. The Nvidia driver has to support CUDA 11.3 (see nvidia-smi command).
I'm sorry, no Windows or Mac support as of yet.

This plugin depends on a number of softwares so there's no easy way to install using plugin managers.
The easiest way to get started is using Docker.

You need to install Nvidia-docker. From their official guide,

curl https://get.docker.com | sh \
  && sudo systemctl --now enable docker

distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
      && curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
      && curl -s -L https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.list | \
            sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
            sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

sudo apt-get update
sudo apt-get install -y nvidia-docker2
sudo systemctl restart docker

Warning: this will download around 7GB. This program has many dependencies that makes it huge.

git clone https://github.com/kiyoon/nvim-hand-gesture
docker pull kiyoon/nvim-hand-gesture

# You may need to change the variables
WEBCAM=/dev/video0
REPO_PATH=~/nvim-hand-gesture
docker run --gpus all --rm -it \
    --device=$WEBCAM:/dev/video0 \
    -v "$REPO_PATH":/workspace kiyoon/nvim-hand-gesture

To enable the GUI demo,

xhost +local:docker
docker run --gpus all --rm -it \
    --device=$WEBCAM:/dev/video0 \
    -e DISPLAY=${DISPLAY} -v /tmp/.X11-unix:/tmp/.X11-unix \
    -v "$REPO_PATH":/workspace kiyoon/nvim-hand-gesture --gui

" With GUI demo
nmap <leader>G <Cmd>call system("docker run --gpus all --rm --device=/dev/video0:/dev/video0 -e DISPLAY=${DISPLAY} -v /tmp/.X11-unix:/tmp/.X11-unix -v ~/project/nvim-hand-gesture:/workspace -v /run/user:/run/user kiyoon/nvim-hand-gesture --gui --nvim_socket_path " . v:servername . " &")<CR>
" Without GUI
nmap <leader>g <Cmd>call system("docker run --gpus all --rm --device=/dev/video0:/dev/video0 -v ~/project/nvim-hand-gesture:/workspace -v /run/user:/run/user kiyoon/nvim-hand-gesture --nvim_socket_path " . v:servername . " &")<CR>
" Quit running process
nmap <leader><leader>g <Cmd>let g:quit_nvim_hand_gesture = 1<CR>

Note that you need to remove -it option, and may change the webcam device and nvim-hand-gesture repo path to the correct one.
-v /run/user mount is because Neovim's v:servername is usually /run/user/$USER/nvim.$pid.0. It may depend on the system.

Change lua files in gesture_mappings to your likings. If the action is in persistent_gestures.txt, the lua script will keep being executed until you stop that action. If not, it will only be executed once and be ignored for the rest of the duration.

You may not want to make a gesture command intrusive to your original neovim workflow. To make it as practical as possible, try to avoid moving your cursor or amending text.

Instead, some example use cases include:

• Opening / closing a side bar (e.g. nvim-tree) without focussing on it
• Opening a floating window (e.g. peek function definition).
• Adjusting size of splits

I'd be happy if anyone shares their own use cases and good sets of defaults!

This is a porting of TSM demo in order to use it to control Neovim.