DeepLabCut WebUI

Docs

For more information on how to use this software, check out the online documentation. https://jcwillox.github.io/deeplabcut-webui/docs

Setup

Requirements

• node, we are targeting the latest LTS version, currently NodeJS 16.
• pnpm, is the frontend package manager, just enable corepack using corepack enable and it will automatically be installed when you try to use it. See pnpm.io/installation#using-corepack for more info.
• task, this is optional but is needed for running the scripts in Taskfile.yaml, see taskfile.dev/#/installation. For Windows users we have a script to download task and add it to the system path, see scripts/install-task.ps1.

Using a Virtual Environment

• Install Python 3.8+, on debian you may also need to install the python3-venv package.
• Install ffmpeg, if you are on Windows and used the install-task.ps1 script you can download ffmpeg-full from gyan.dev and place the exe files in ~/.local/bin.
• Clone and cd into the repository.
• Run task setup, this will create a Python virtual environment in .venv and install the frontend and backend dependencies.
  • This can also be run to update the environments/dependencies after pulling the project.
• Everything is set up now, however, you will need to change the Python interpreter inside your editor manually to use the virtual environment.
  • VSCode: open the backend/app/main.py file, press F1, find "Python: Select interpreter", select the interpreter called .venv, you may need to click the refresh icon in the top-right first.
• You can start the frontend and backend with task fullstack, run task -l to see additional scripts.
• Additionally, when using VSCode you can press F1 and search for Task: Run Task, then select whether you'd like to run the frontend, backend or both.

# after installing node and python
corepack enable
git clone https://github.com/jcwillox/deeplabcut-webui
cd deeplabcut-webui
task setup
# run frontend & backend
task fullstack

Using Conda (inc. DeepLabCut)

• Install conda, it's recommended to install miniconda. Make sure you tell it to add conda to your PATH.
• GPU Support (optional) you will need to install your graphics card driver (you likely already have) and then install cuda support https://developer.nvidia.com/cuda-downloads. See here for more info.
• After installing conda run conda init <shell>, e.g. conda init powershell.
  • You can revert this later with conda init <shell> --reverse.
• Optional: Run conda config --set auto_activate_base false this will stop conda from switching to the default conda (base) environment whenever you open PowerShell.
• Clone and cd into the repository.
• Run task setup-conda-node, this will set up the environments/dependencies for the project.
  • This can also be run to update the environments/dependencies after pulling the project.
• Run conda activate dlc-webui.
• Everything is set up now, however, you will need to change the Python interpreter inside your editor manually to use the conda environment.
  • VSCode: open the backend/app/main.py file, press F1, find "Python: Select interpreter", select the interpreter called dlc-webui, you may need to click the refresh icon in the top-right first.
• You can start the DeepLabCut GUI with task dlc-gui, or start the frontend and backend with task fullstack, run task -l to see additional scripts.
• Additionally, when using VSCode you can press F1 and search for Task: Run Task, then select whether you'd like to run the frontend, backend or both.

# after installing node and conda
corepack enable
git clone https://github.com/jcwillox/deeplabcut-webui
cd deeplabcut-webui
task setup-conda-node
# run frontend & backend
conda activate dlc-webui
task fullstack