The ModelNet dataset contains 662 target categories, 127,915 CAD models, and 10 categories of oriented data. It aims to provide comprehensive 3D object models for researchers in computer vision, computer graphics, robotics, and cognitive science. The dataset consists of three subsets:

• ModelNet10 is a subset of data with 10 labeled orientations;
• ModelNet40 is a 3D model of 40 categories;
• Aligned40 is a 3D model labeled with 40 categories. The ModelNet dataset was released by the Princeton Vision and Robotics Laboratory in 2015. The main publishers are N. Sedaghat, M. Zolfaghari, E. Amiri and T. Brox. The related paper is "3D ShapeNets: A Deep Representation for Volumetric Shapes"