Common Objects in 3D is a multi-view image dataset of common objects. The images are annotated with camera angles and ground truth 3D point clouds. The dataset contains 1.5 million frames from nearly 19,000 videos. These videos capture objects from 50 MS-COCO categories. Therefore, it outperforms other datasets in terms of both the number of categories and objects.

This dataset can be used to learn and evaluate real-life 3D class reconstruction at scale.