KITTI-360

A two-stage baseline implemented by KITTI-360 authors. Appearance
synthesis is performed by Free View Synthesis using the released model
trained on the Tanks and Temples dataset. We then apply PSPNet on the
synthesized images for semantic segmentation.

none

@InProceedings{Riegler2020ECCV
author = {Gernot Riegler and Vladlen Koltun},
title = {Free View Synthesis},
booktitle = {ECCV},
year = {2020},
},
@InProceedings{Zhao2017CVPR,
author = {Hengshuang Zhao and Jianping Shi
and Xiaojuan Qi and Xiaogang Wang and Jiaya Jia},
title = {Pyramid Scene Parsing Network},
booktitle = {CVPR},
year = {2017},
}