Compositional Human Pose Regression
This work addresses the challenge of improving regression-based methods for human pose estimation, which is important for applications like computer vision and robotics, though it is incremental in nature.
The paper tackles the problem of human pose estimation by proposing a structure-aware regression approach that uses a reparameterized pose representation with bones and a compositional loss function to exploit joint connections, achieving state-of-the-art results on Human3.6M and competitive performance on MPII.
Regression based methods are not performing as well as detection based methods for human pose estimation. A central problem is that the structural information in the pose is not well exploited in the previous regression methods. In this work, we propose a structure-aware regression approach. It adopts a reparameterized pose representation using bones instead of joints. It exploits the joint connection structure to define a compositional loss function that encodes the long range interactions in the pose. It is simple, effective, and general for both 2D and 3D pose estimation in a unified setting. Comprehensive evaluation validates the effectiveness of our approach. It significantly advances the state-of-the-art on Human3.6M and is competitive with state-of-the-art results on MPII.