Fig. 2: Visualization of 2D Pose Estimation Process.
From: Learning aggressive animal locomotion skills for quadrupedal robots solely from monocular videos
a Images from the source video. b 2D pose estimation by DLC Superanimal quadruped model. c Refined skeleton with joints in pink and bones in white. d 2D pose visualization of joint position estimates directly from the DLC Superanimal quadruped model. The reconstructed trajectories are discontinuous and cannot be used for learning. e Joint trajectories are processed by our proposed method. It can be clearly seen that our reconstructed joint trajectories are significantly smoother compared to the DLC Superanimal quadruped model. f Visualization of the reconstructed joint trajectories. (Image source: YouTube. My 4 flipping dogs).
