is an open-source Python toolbox designed for high-performance, markerless pose estimation of animals and objects using deep learning. By leveraging transfer learning, the software allows researchers to achieve human-level tracking accuracy with remarkably little training data—typically only 50 to 200 manually labeled frames. It features a user-friendly graphical interface (GUI) and support for both 2D and 3D tracking across various species and behaviors, making it a standard tool in fields like neuroscience, ethology, and biomechanics.
