This is pytorch code of the paper "Semi-supervised Bidirectional Alignment for Remote Sensing Cross-domain Scene Classification" of ISPRS-2023. If it is helpful, please kindly cite it as:
@article{HUANG2023192,
title = {Semi-supervised bidirectional alignment for Remote Sensing cross-domain scene classification},
journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
volume = {195},
pages = {192-203},
year = {2023},
issn = {0924-2716},
doi = {https://doi.org/10.1016/j.isprsjprs.2022.11.013},
url = {https://www.sciencedirect.com/science/article/pii/S0924271622003069},
author = {Wei Huang and Yilei Shi and Zhitong Xiong and Qi Wang and Xiao Xiang Zhu},
}
Please separate the dataset into source set, labeled target set, and unlabeled target set.
If you want to train the model, for example, please run:
CUDA_VISIBLE_DEVICES=7 python main.py --method BSCA --net resnet34 --source WHU --target RSSCN7 --steps 2001 --thr 0.5 --sample_per_class 3 --save_check