Pytorch implementation of the paper "Affinity Space Adaptation for Semantic Segmentation Across Domains", TIP, 2020. In this paper, we address the problem of unsupervised domain adaptation (UDA) in semantic segmentation, achieving the state-of-the-art performance on standard benchmarks.
If you find this paper useful in your research, please consider citing:
@ARTICLE{9184275,
author={W. {Zhou} and Y.{Wang} and J. {Chu} and J. {Yang} and X. {Bai} and Y. {Xu}},
journal={IEEE Transactions on Image Processing},
title={Affinity Space Adaptation for Semantic Segmentation Across Domains},
year={2020},
volume={},
number={},
pages={1-1},}
- Comparison Results on Cityscapes when adapted from GTA5 in terms of per-class IoU and mIoU over 19 class.
- Comparison Results on Cityscapes when adapted from SYTNTHIA in terms of per-class IoU and mIoU over 13 or 16 class.
- Download the GTA5 Dataset as source dataset.
- Download the Cityscapes Dataset as target dataset.
Initial weights and trained models can be downloaded from here. [Google Drive] [Baidu Drive (download code: 9lov) ].
Put the weights in the "ASANet/pretrained" directory.
bash scripts/train_gta2city.sh
bash scripts/evaluate.sh
Thanks for your attention! If you have any suggestion or question, you can leave a message here or contact us directly:
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent