DeepLabCut Networks for 2D Pose Estimation

This folder contains trained DeepLabCut networks for 2D pose estimation from multiple viewpoints, as described in this work. It includes three subfolders, each corresponding to a different network for performing 2D pose estimation on side (cam1), side-front (cam2), and front (cam3) views. We used the same networks for mirror cameras (e.g., the right and left sides). For more details on how to use these networks for 2D pose estimation and how to triangulate them to obtain 3D pose estimation, please visit the NeLy-EPFL/kinematics3d GitHub repository. The accompanying PDF (tracked_keypoints_setup.pdf) provides information about the locations of the tracked points. To uncompress the file, run the following command in the terminal: $ tar -xvf filename.tgz