ModelNet10 Princeton 3D Object Dataset

The goal of the Princeton ModelNet project is to provide a comprehensive, legible collection of 3D CAD models of objects to researchers in the fields of computer vision, computer graphics, robotics, and cognitive science.

content

The ModelNet10 dataset is part of the ModelNet40 dataset and contains 4,899 pre-aligned shapes of 10 types of CAD furniture models, such as bathtubs, beds, chairs, tables, etc. 3,991 (80%) shapes are used for training and 908 (20%) shapes are used for testing. The CAD models are in object file format (OFF). Matlab functions for reading and visualizing OFF files are provided in the Princeton Vision Toolkit (PVT).

To build the core of the dataset, the research team compiled a list of the most common object categories in the world using statistics obtained from the SUN database. After building the object vocabulary, an online search engine was used to collect 3D CAD models belonging to each object category by querying each object category term. Then, human workers on Amazon Mechanical Turk were hired to manually judge whether each CAD model belongs to the specified category using an in-house designed tool with quality control capabilities. In order to obtain a very clean dataset, 10 popular object categories were selected, and models that did not belong to these categories were manually removed. In addition, manual alignment of CAD model orientations was performed on a subset of 10 categories.