Abstract: Fusion of features from multi-view is one of the effective means to improve multi-view 3D pose estimation. It relies on robust and accurate 2D pose estimation, and effective fusion methods.