neuroinformatics-unit/movement

What it solves

Deep learning tools like DeepLabCut and SLEAP can track animal movements in videos with high accuracy, but there is no standardised way to process the resulting tracks. Researchers often need to clean, visualise, and quantify these tracks before they can use them in their studies. movement fills this gap by providing a consistent, modular interface for analysing motion tracks across different tools and file formats.

How it works

movement offers a set of Python functions and data structures designed to load, clean, visualise, and quantify animal movement tracks. It is built to support popular tracking frameworks and file formats, so users can work with data from DeepLabCut, SLEAP, and others without rewriting their analysis pipelines. The toolbox also integrates with the napari viewer, giving users an interactive graphical interface for exploring and inspecting their tracked data.

Who it’s for

This toolbox is aimed at researchers and scientists who use deep learning-based motion tracking in fields such as neuroscience, biomechanics, conservation, and ethology. It is for anyone who needs a reliable and standardised way to process animal position data after tracking, regardless of which tracking tool they originally used.

Highlights

• Provides a uniform interface for handling motion tracks from multiple tracking frameworks
• Includes tools for data cleaning, visualisation, and motion quantification
• Supports popular file formats used by DeepLabCut and SLEAP
• Offers a graphical user interface through napari for interactive data exploration
• Designed as an open-source, community-driven project with contributions from the neuroinformatics unit