Thanks for your interest in our work, "HFA-Net:High Frequency Attention Siamese Network for Building Change Detection in VHR Remote Sensing Images". More detailed information can be find at DOI: 10.1016/j.patcog.2022.108717
If you find our work useful for your research, please consider citing our paper:
@article{ZHENG2022108717,
title = {HFA-Net: High frequency attention siamese network for building change detection in VHR remote sensing images},
journal = {Pattern Recognition},
volume = {129},
pages = {108717},
year = {2022},
issn = {0031-3203},
}
Firstly, you can arrange your datasets as follows:
├── YourDataset
│ ├── train
│ │ ├── A
│ │ ├── B
│ │ └── label
│ ├── test
│ │ ├── A
│ │ ├── B
│ │ └── label
............
Then you can set the path to the dataset in Dataset.py. It should be noted that all the images need to be named as "number.tif", e.g., "1.tif,11.tif,111.tif".
The batch sizes for training and testing, optimizers, learning rate and so on can be set or changed in Main.py.
After the corresponding environment is installed successfully, excute the code below to train and test HFA-Net over your dataset.
python Main.py --dataset $YourDataset
The evaluation metrics and the best visualized results will be saved for your convenience.
pytorch 1.8.0 with corresponding CUDA toolkits.
torchvision 0.9.0
argparse 1.4.0
opencv-python 4.5.4.58
tqdm 4.62.3
If there are unavoidable problems or inconveniences for you to directly implement HFA-Net, you can easily extract the bare untrained model of HFA-Net saved in Net.py, Attention_Module.py, High_Frequency_Module.py, Encoder.py and Decoder.py and utilize it in your work for comparison.
In addition, you can directly acquire the quantitative experimental results of HFA-Net over three widely used VHR change detection data sets.
We are grateful for outstanding contributions of three open change detection data sets, i.e., WHU-CD [1], LEVIR-CD [2], Google Data set [3]:
[1] Ji S, Wei S, Lu M. Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 57(1): 574-586.
[2] Chen H, Shi Z. A spatial-temporal attention-based method and a new dataset for remote sensing image change detection[J]. Remote Sensing, 2020, 12(10): 1662.
[3] D. Peng, L. Bruzzone, Y. Zhang, H. Guan, H. Ding and X. Huang. SemiCDNet: A Semisupervised Convolutional Neural Network for Change Detection in High Resolution Remote-Sensing Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(7): 5891-5906.