Installation

Installation#

This page explains how to install DeepLabCut-live-GUI for interactive, real‑time pose estimation.

We support various installation methods, including uv and mamba/conda.

Tip

If you feel confident you meet the requirements and you just want to get started quickly, see the Quickstart (recommended defaults) section below.

System requirements#

Key takeaways#

  • On Windows: If you want TensorFlow, use Python 3.10

  • On macOS: TensorFlow is only supported on CPU

  • On Linux: Full support for both PyTorch and TensorFlow

OS support#

OS

PyTorch

TensorFlow

Notes & recommendations

Windows

✅

❌

Limited TensorFlow support due to lack of official Windows builds for Python 3.11+ onwards

Linux

✅

✅

Full support for both backends

macOS

✅

⚠️

PyTorch MPS support is improving but still has limitations; TensorFlow only supports CPU on macOS

Hardware requirements#

  • Any compatible camera (see Camera support):

    • USB webcam, OBS virtual camera → OpenCV-recognized cameras are accessible by default

    • Basler

    • GenTL [1]

    • Aravis [1]

  • Optional but recommended:

    • CUDA-capable GPU (for real‑time inference)

    • NVIDIA drivers compatible with your PyTorch/TensorFlow version

Note

If you use an OpenCV-compatible camera (e.g. USB webcam, OBS virtual camera), you can simply install the package as it comes with OpenCV support by default.

Software requirements#

  • Python 3.10, 3.11 or 3.12

  • A machine learning framework for inference (instructions below for both):

    • PyTorch (recommended for best performance and compatibility)

    • TensorFlow (for backwards compatibility with existing models)

  • A working camera backend (see Camera support)

Choose your installation method#

Below instructions cover installation with uv and mamba/conda, but you may also install with other package managers like pdm if preferred.

Note

The main DeepLabCut package and its GUI are not required to use this software, as it is designed to be a lightweight interface for real‑time pose estimation.

Install DeepLabCut-live-GUI#

Danger

This beta version of the package is not currently on PyPI. For this reason, the install process requires cloning the repository and installing from source.

Clone the repository#

git clone https://github.com/DeepLabCut/DeepLabCut-live-GUI.git
cd DeepLabCut-live-GUI

Install with uv#

We recommend installing with uv, but also support installation with pip or conda (see next section).

Create and activate a new virtual environment#
uv venv -p 3.12 # or desired Python version
source .venv/bin/activate

uv venv -p 3.12 # or desired Python version
.\.venv\Scripts\activate.bat

uv venv -p 3.12 # or desired Python version
.\.venv\Scripts\Activate.ps1
Choose inference backend#

We offer two distinct inference backends: PyTorch and TensorFlow. You may install either or both, but you must choose at least one to run the pose estimation models.

Important

To enable GPU support and obtain detailed installation instructions, please refer to the official PyTorch installation guide and install PyTorch before installing the GUI package.

uv pip install -e ".[pytorch]"

Caution

Please note TensorFlow is no longer available on Windows for Python > 3.10.

uv pip install -e ".[tf]"

Note

For detailed installation instructions, please refer to the official TensorFlow installation guide.

Install with mamba or conda#

Create and activate a new conda environment#

If you prefer using mamba or conda, you can create a new environment and install the package with:

conda create -n dlclivegui python=3.12 # pick your desired Python version
conda activate dlclivegui
Choose inference backend#

We offer two distinct inference backends: PyTorch and TensorFlow. You may install either or both, but you must choose at least one to run the pose estimation models.

Important

To enable GPU support and obtain detailed installation instructions, please refer to the official PyTorch installation guide and install PyTorch before installing the GUI package.

pip install -e ".[pytorch]"

Caution

Please note TensorFlow is no longer available on Windows for Python > 3.10.

pip install -e ".[tf]"

Note

For detailed installation instructions, please refer to the official TensorFlow installation guide.

Verify installation#

After installation, you can verify that the package is installed correctly with:

dlclivegui --help

Download and export a model from the model zoo#

See the Pre-trained models page for instructions on how to programmatically download and export pre-trained models from the DeepLabCut Model Zoo for use in the GUI.

Important

We may in the future add more direct, built-in support for browsing and downloading compatible models. For now, you can use the dlclive.modelzoo API to fetch and export models as described in the linked documentation.

Run the application#

After installation, you can start the DeepLabCut-live-GUI application with:

dlclivegui # OR uv run dlclivegui

Important

Make sure your venv or conda environment is activated before running the application, so it can access the installed dependencies.

Contents