poseinterface

poseinterface exists to advance pose estimation and point tracking applications in animal behaviour videos. The project aims to:

Build a benchmark dataset with dozens of videos and annotations from multiple institutes, open to external contributions.
Develop a general-purpose framework for running pose estimation and point tracking tools on benchmark data, and for comparing their outputs via evaluation metrics.
Provide baseline models trained using common pose estimation and tracking frameworks on the benchmark datasets.

Read the documentation for more information.

Warning

This project is in early stages of development. The API is not stable and may change without warning. Use with caution.

Installation

We recommend installing poseinterface in a virtual environment, using uv.

In your working directory, create a new environment and activate it:

uv venv --python=3.13
source .venv/bin/activate  # On macOS and Linux
.venv\Scripts\activate     # On Windows PowerShell

Then, install the package directly from the main branch on GitHub:

uv pip install git+https://github.com/neuroinformatics-unit/poseinterface.git@main

If you would like to contribute, see the contributing guide, which includes instructions for setting up a development environment.

This package layout and configuration (including pre-commit hooks and GitHub actions) have been copied from the python-cookiecutter template.