Pix3D is a single-image 3D shape modeling dataset and method, a large-scale benchmark of image-shape pairs with pixel-level 2D-3D alignment, which is widely used in shape-related tasks including reconstruction, retrieval, viewpoint estimation, etc.

The dataset was jointly released by the Department of Computer Science at Princeton University and the Massachusetts Institute of Technology. The main publishers are Xingyuan Sun, Jiajun Wu, Xiuming Zhang, Zhoutong Zhang, Chengkai Zhang, Tianfan Xue, Joshua B. Tenenbaum and William T. Freeman.