lambdaloop/anipose
An open-source toolkit for markerless 3D pose estimation of animal behavior using multi-camera views and deep learning.

Velocity · 7d
+0.2
★ / day
Trend
→steady
star history
Anipose estimates 3D poses of animals from video without physical markers by combining 2D keypoint detection from DeepLabCut with geometric triangulation across synchronized camera views. It is designed for behavioral neuroscience and biomechanics research, providing calibration and filtering utilities to produce reliable 3D trajectories from laboratory video recordings.