A markerless pose estimator applicable to limbless animals.
The analysis of kinematics, locomotion, and spatial tasks relies on the accurate detection of animal position and pose. Most available trackers represent animals as single points (no pose) or rely on markers — physical (paint, stickers, beads) or in silico (joints, limbs, colour patterns) — to build a skeletal representation. Markers often cannot be used on small animals, on animals lacking prominent body structures, or in environments (e.g. aquatic) that detach them.
LACE builds the pose of an animal de novo from its contour: it detects the contour, derives the body mid-line, and constructs a pseudo-skeleton of vertices and edges. Applied to larval Drosophila melanogaster and adult zebrafish, LACE quantifies, for example, genetic alterations of peristaltic movements and sex-specific locomotion patterns associated with different body shapes — even in animals without limbs.
LACE is the original MATLAB implementation. The method has since been ported to Python and is actively maintained there:
- pyLACE — Python port of the LACE tracker.
- pyLACEpostHoc — database, post-hoc analysis, and plotting layer for the LACE family.
New work should generally start from the Python successors; this repository is preserved as the reference implementation accompanying the original publication.
- MATLAB R2021a or later (R2024b used in CI)
- Image Processing Toolbox
Toolbox (recommended): download LACE.mltbx from the
latest release and
double-click it to install via the MATLAB Add-On Manager.
From source: clone the repository and add it to the MATLAB path:
addpath(genpath('LACE'));
savepath;The GUIDE apps in GUI/ are the primary entry points:
ET_GUI_Spline— spline / contour tracing and background handlingET_GUI_ana_Fish,ET_GUI_ana_Fly— species-specific analysis GUIsET_GUI_Benzer— Benzer-assay scripting
See Contents.m for the full module map (help LACE).
The repository uses the MATLAB build tool (buildfile.m):
buildtool check % static code analysis
buildtool test % run the unit-test suite (tests/)
buildtool toolbox % package release/LACE.mltbxThe same tasks run in CI through the
MathWorks matlab-actions suite.
If you use LACE, please cite:
Garg, V., André, S., Giraldo, D., Heyer, L., Göpfert, M. C., Dosch, R., & Geurten, B. R. H. (2022). A markerless pose estimator applicable to limbless animals. Frontiers in Behavioral Neuroscience, 16, 819146. https://doi.org/10.3389/fnbeh.2022.819146
A machine-readable citation is provided in CITATION.cff.
GPL-3.0-or-later — see LICENSE.