SemanticHouseCat3D

SemanticHouseCat3D is a synthetic dataset of common household objects designed for benchmarking pose-aware and 3D-consistent semantic matching. It includes objects from six categories of the HouseCat6D dataset and precise keypoint correspondences annotations are made from an automatic procedure with minor manual refinements. This dataset is ideal for evaluating and developing new methods targeting robust object manipulation, affordance detection, and 3D-aware matching in robotics and computer vision.