A Markerless Pose Estimator Applicable to Limbless Animals.

Area of Science:

• Animal behavior and biomechanics
• Computational biology and computer vision
• Zoology and developmental biology

Background:

• Accurate animal position and pose detection is crucial for analyzing kinematics, locomotion, and spatial behaviors.
• Existing video analysis trackers often rely on single-point representations or physical/in silico markers, which have limitations for small, limbless, or aquatic animals.
• The constraints of current tracking methods necessitate innovative approaches for comprehensive pose estimation.

Purpose of the Study:

• To introduce LACE (Limbless Animal traCker), a novel marker-free pose-estimation tool.
• To enable detailed kinematic and locomotion analysis using only the animal's contour.
• To provide a versatile method for assessing movement patterns in diverse animal models, including those lacking limbs.

Main Methods:

• LACE detects animal contours and derives a body midline to construct a pseudo-skeleton.
• The method defines vertices and edges from the contour to represent pose.
• The approach was validated using larval *Drosophila melanogaster* and adult zebrafish.

Main Results:

• LACE successfully estimates pose and position from animal contours without physical or in silico markers.
• The method allows quantification of genetic effects on peristaltic movements in larval *Drosophila melanogaster*.
• LACE revealed gender-specific locomotion patterns correlated with body shape variations in adult zebrafish.

Conclusions:

• LACE offers a versatile and marker-free solution for animal pose and position tracking.
• The method is effective for analyzing complex movements and morphological variations in diverse species.
• This approach advances the study of animal locomotion and behavior, particularly for challenging subjects.