ModelNet-R is a refined version of the widely used ModelNet40 benchmark for 3D point cloud classification. It applies systematic refinements and robustness transformations to improve evaluation reliability in real-world scenarios. The dataset contains 3D CAD models represented as point clouds, labeled across 40 object categories, and is designed for benchmarking classification methods in point cloud research.
