Awesome

CSA-CDGAN: Channel Self-Attention Based Generative Adversarial Network for Change Detection of Remote Sensing Images

A general framework for change detection of remote sensing images

Requirements

Python 3.7.0  
Pytorch 1.6.0  
Visdom 0.1.8.9  
Torchvision 0.7.0

Datasets

• CDD dataset
  • paper: [Change detection in remote sensing images using conditional adversarial networks. International Archives of the Photogrammetry](https://drive.google.com/file/d/1GX656JqqOyBi_Ef0w65kDGV to-nHrNs9)
  • WHU-CD: Fully Convolutional Networks for Multisource Building Extraction From an Open Aerial and Satellite Imagery Data Set
  • LEVIR-CD: A Spatial-Temporal Attention-Based Method and a New Dataset for Remote Sensing Image Change Detection
    You also can download datasets after being processed by us. [Baiduyun] the password is hnbi. or [GoogleDrive]

Pretrained Model

Pretrained models for CDD, LEVIR-CD and WHU-CD are available. You can download them from the following link. [Baiduyun] the password is yudl. [GoogleDrive]

Test

Before test, please download datasets and pretrained models. Revise the data-path in constants.py to your path. Copy pretrained models to folder './dataset_name/outputs/best_weights', and run the following command:

cd CSA-CDGAN_ROOT
python make_dataset.py
python test.py

make_dataset.py can generate .txt files for training, validation and test. Not that the dataset structure should be the same as following:

Custom dataset
|--train
  |--file1
    |--t0.jpg, t1.jpt, label.jpg
  |--file2
    |--t0.jpg, t1.jpt, label.jpg
  ...
|--test
  |--file1
    |--t0.jpg, t1.jpt, label.jpg
  |--file2
    |--t0.jpg, t1.jpt, label.jpg
  ...
|--validation
  |--file1
    |--t0.jpg, t1.jpt, label.jpg
  |--file2
    |--t0.jpg, t1.jpt, label.jpg
  ...

Training

cd CSA-CDGAN_ROOT
python make_dataset.py
python -m visdom.server
python train.py

To display training processing, open 'http://localhost:8097' in your browser.

Citing TransCD

If you use this repository or would like to refer the paper, please use the following BibTex entry.

@inproceddings{TransCD,
title={CSA-CDGAN: Channel Self-Attention Based Generative Adversarial Network for Change Detection of Remote Sensing Images},
author={ZHIXUE WANG, YU ZHANG*, LIN LUO, NAN WANG},
yera={2021},
}

Reference

-Akcay, Samet, Amir Atapour-Abarghouei, and Toby P. Breckon. "Ganomaly: Semi-supervised anomaly detection via adversarial training." Asian conference on computer vision. Springer, Cham, 2018.