
Diffusion models with multi-modal inputs effectively handle symmetry and uncertainty in category-level pose estimation.


coremsg

diffusionnocs-managing-symmetry-and-uncertainty-in-category-level-pose-estimation


DiffusionNOCS: Managing Symmetry and Uncertainty in Category-level Pose Estimation


title_rewrite


提案されたMRC-Netは、単一のRGB画像から利用可能な3D CADモデルを使用してオブジェクトの6DoFポーズを決定するシングルショットアプローチです。


mrc-net-6-dof-pose-estimation-with-multiscale-residual-correlation


MRC-Net: 6-DoF Pose Estimation with MultiScale Residual Correlation



Proposing a probabilistic model using diffusion to estimate dense canonical maps for recovering partial object shapes and establishing correspondences essential for pose estimation.
