OPA stands for Object Placement Assessment. It is an object placement dataset based on the COCO dataset. It provides a large number of synthetic images and binary labels of the rationality of the placement of foreground objects, covering 47 foreground categories and a total of 73,470 synthetic images. The placement of foreground objects in positive synthetic images is reasonable, while the placement of foreground objects in negative synthetic images is unreasonable. The dataset also provides the foreground objects and background images that make up the synthetic images.

Citation

@article{liu2021OPA,
title={OPA: Object Placement Assessment Dataset},
author={Liu,Liu and Liu,Zhenchen and Zhang,Bo and Li,Jiangtong and Niu,Li and Liu,Qingyang and Zhang,Liqing},
journal={arXiv preprint arXiv:2107.01889},
year={2021}
}