ASFR-Net: Adversarial Alignment and Spatio-Frequency Refinement Network for Heterogeneous Remote Sensing Image Change Detection

Authors: Xin-Jie Wu, Zhi-Hui You, Si-Bao Chen, Qing-Ling Shu, Xiao Wang, Jin Tang, and Bin Luo

This repository contains the official PyTorch implementation of our paper ASFR-Net.

1. Overview

ASFR-Net is an end-to-end adversarial spatio-frequency refinement network. It bridges the modality gap via a Modality-Invariant Representation Learner (MIR-Learner), suppresses sensor-specific noise using a Spatio-Frequency Synergistic Enhancement Module (SFEM), and generates precise change maps through a decoder equipped with Hierarchical Guided Fusion Module (HGFM) blocks.

2. Usage

• Prepare the dataset:

  • Download our VisNIR-HCD dataset: Google Drive
  • Organize the dataset directory as follows:

  ├─VisNIR-HCD
      ├─label       ...png (Change masks)
      ├─list        ...txt (Train/Val/Test splits)
      ├─NIR_A       ...png (NIR images at T1/T2)
      ├─NIR_B       ...png
      ├─RGB_A       ...png (Visible images at T1/T2)
      └─RGB_B       ...png
  

• Prerequisites:

  • Create a virtual environment: conda create -n ASFRNet python=3.8
  • Install dependencies: pip install -r requirements.txt

• Clone this repo:

  git clone https://github.com/LuoYang2024/ASFR-Net.git
  cd ASFR-Net

• Train/Test:

  • To train the model: python ./tools/train.py
  • To test the model: python ./tools/test.py

3. Change Detection Results

Quantitative comparisons on VisNIR-HCD, MT-Wuhan, and XiongAn datasets. ASFR-Net consistently outperforms existing SOTA methods in F1-score and IoU.

4. Checkpoints on Datasets

Our trained pth on VisNIR_HCD: Download

Our trained pth on MT_Wuhan: Download

Our trained pth on XiongAn: Download

5. Acknowledgment

This repository is built under the help of the projects A2Net and RFANet for academic use only.

6. Citation

Please cite our paper if you find this work or the dataset useful:

@article{Wu2026_ASFRNet,
  title={ASFR-Net: Adversarial Alignment and Spatio-Frequency Refinement Network for Heterogeneous Remote Sensing Image Change Detection},
  author={Wu, Xin-Jie and You, Zhi-Hui and Chen, Si-Bao and Shu, Qing-Ling and Wang, Xiao and Tang, Jin and Luo, Bin},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  year={2026},
  doi={10.1109/TGRS.2026.xxxxxxx}
}