ShapeNet数据集

ShapeNet是一个由对象的三维CAD模型表示的丰富注释的，大规模的形状存储库。ShapeNet包含来自多种语义类别的3D模型，并按照WordNet分类法组织它们。它是一组数据集，为每个3D模型提供许多语义标注，如一致的刚性对准、零件和双边对称平面、物理尺寸、关键字以及其他计划的标注。注释通过基于web的公共接口提供，以支持对象属性的数据可视化，促进数据驱动的几何分析，并为计算机图形学和视觉研究提供大规模定量基准。ShapeNet已经索引了超过300万个模型，其中22万个模型被分为3135个类别(WordNet synsets)。

ShapeNet is a large-scale, richly annotated shape repository built from 3D CAD models of real-world objects. It houses 3D models across a diverse array of semantic categories, which are organized following the WordNet taxonomy. As a collection of datasets, ShapeNet offers multiple semantic annotations for each 3D model, such as consistent rigid alignments, part and bilateral symmetry planes, physical dimensions, keywords, and other planned annotations. These annotations are accessible through a web-based public interface, enabling data visualization of object attributes, facilitating data-driven geometric analysis, and providing large-scale quantitative benchmarks for computer graphics and computer vision research. To date, ShapeNet has indexed more than 3 million 3D models, out of which 220,000 models have been categorized into 3,135 WordNet synsets.

搜集汇总

背景与挑战

背景概述

ShapeNet数据集是一个大规模、丰富注释的3D形状存储库，包含超过300万个3D模型，其中22万个模型被分类为3135个语义类别，按照WordNet分类法组织。该数据集提供多种语义标注，如刚性对准、零件信息和物理尺寸，旨在支持计算机图形学和视觉研究中的数据可视化、几何分析和基准测试。数据量为76.78GB，由成都索贝数码科技股份有限公司创建，属于国家重点研发计划项目。