Stable Neighbor Denoising for Source-free Domain Adaptive Segmentation (CVPR-2024)
Python 3.8.0
pytorch 1.10.1
torchvision 0.11.2
einops 0.3.2
Please see requirements.txt for all the other requirements.
Data:
- Download The Cityscapes Dataset
The data folder should be structured as follows:
├── datasets/
│ ├── cityscapes/
| | ├── gtFine/
| | ├── leftImg8bit/
...
Pretrain models:
- Download pretrained model on GTA5: (GTA5_DG)
- Download pretrained model on (SYNTHIA_DG) Then, put these *.pth into the pretrain folder.
G2C-warm:
CUDA_VISIBLE_DEVICES=0 nohup python train_SND.py -cfg configs/deeplabv2_r101_dtst_G2C.yaml OUTPUT_DIR results/G2C_SND_WARM/ resume pretrain/G2C_model_iter020000.pth > logs/G2C_SND_WARM.file 2>&1 &
cp -r results/G2C_SND_WARM/ results/G2C_SND/
SND:
CUDA_VISIBLE_DEVICES=3 nohup python train_SND.py -cfg configs/deeplabv2_r101_SND_G2C.yaml OUTPUT_DIR results/G2C_SND/ resume pretrain/G2C_model_iter020000.pth > logs/G2C_SND.file 2>&1 &\
Test:
CUDA_VISIBLE_DEVICES=1 python test.py -cfg configs/deeplabv2_r101_SND_G2C.yaml resume results/G2C_SND/
Some codes are adapted from higher We thank them for their excellent projects.
If you find this code useful please consider citing
@inproceedings{zhao2023towards,
title={Towards Better Stability and Adaptability: Improve Online Self-Training for Model Adaptation in Semantic Segmentation},
author={Zhao, Dong and Wang, Shuang and Zang, Qi and Quan, Dou and Ye, Xiutiao and Jiao, Licheng},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={11733--11743},
year={2023}
}
@InProceedings{Zhao_2024_CVPR,
author = {Zhao, Dong and Wang, Shuang and Zang, Qi and Jiao, Licheng and Sebe, Nicu and Zhong, Zhun},
title = {Stable Neighbor Denoising for Source-free Domain Adaptive Segmentation},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2024},
pages = {23416-23427}
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent