KITTI-360

A two-stage baseline implemented by KITTI-360 authors. Appearance
synthesis is performed by Mip-NeRF. Semantic segmentation predictions
are then obtained by applying a pre-trained PSPNet to the synthesized
images.

None

@article{barron2021mipnerf,
title={Mip-NeRF: A Multiscale Representation
for Anti-Aliasing Neural Radiance Fields},
author={Jonathan T. Barron and Ben Mildenhall and
Matthew Tancik and Peter Hedman and
Ricardo Martin-Brualla and Pratul P. Srinivasan},
journal={ICCV},
year={2021}
},
@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},
}