ShapeNet is a large-scale, richly-annotated repository of 3D models, organized under the WordNet taxonomy. It contains over 3 million 3D models, categorized into 3,135 semantic categories. The dataset includes detailed annotations such as rigid alignments, parts, symmetry planes, physical properties, and semantic information. These annotations enable data-driven geometric analysis, shape recognition, and benchmarking for computer graphics and vision research. ShapeNet provides a web-based interface for searching, viewing, and retrieving models through various modalities, including textual keywords, taxonomy traversal, and similarity search. The dataset is continuously expanding, with new models and annotations being added regularly. ShapeNet aims to serve as a comprehensive knowledge base for representing real-world objects and their semantics. It includes a wide range of annotations, such as language-related, geometric, functional, and physical properties, which are essential for various applications in computer graphics and vision. The dataset is built using a hybrid approach, combining algorithmic predictions with human verification to ensure accuracy and consistency. ShapeNet is designed to support research in shape analysis, recognition, and understanding, and it provides a valuable resource for the computer graphics and vision communities. The dataset is continuously being expanded and improved, with future plans to include more annotations and to enhance the dataset's coverage and functionality.ShapeNet is a large-scale, richly-annotated repository of 3D models, organized under the WordNet taxonomy. It contains over 3 million 3D models, categorized into 3,135 semantic categories. The dataset includes detailed annotations such as rigid alignments, parts, symmetry planes, physical properties, and semantic information. These annotations enable data-driven geometric analysis, shape recognition, and benchmarking for computer graphics and vision research. ShapeNet provides a web-based interface for searching, viewing, and retrieving models through various modalities, including textual keywords, taxonomy traversal, and similarity search. The dataset is continuously expanding, with new models and annotations being added regularly. ShapeNet aims to serve as a comprehensive knowledge base for representing real-world objects and their semantics. It includes a wide range of annotations, such as language-related, geometric, functional, and physical properties, which are essential for various applications in computer graphics and vision. The dataset is built using a hybrid approach, combining algorithmic predictions with human verification to ensure accuracy and consistency. ShapeNet is designed to support research in shape analysis, recognition, and understanding, and it provides a valuable resource for the computer graphics and vision communities. The dataset is continuously being expanded and improved, with future plans to include more annotations and to enhance the dataset's coverage and functionality.