Geometric Alignment Across Shape Variation for Efficient Category-level Object Pose Refinement
A novel architecture that effectively addresses the challenge of shape variations in category-level object pose refinement by integrating learnable affine transformations and a cross-cloud transformation mechanism.
