zmoka-zht / bifa

BiFA: Remote Sensing Image Change Detection with Bitemporal Feature Alignment

     TGRS      PDF     

Introduction

This repository is the code implementation of the paper BiFA: Remote Sensing Image Change Detection with Bitemporal Feature Alignment

The current branch has been tested on Linux system, PyTorch 1.12.0 and CUDA 12.1, supports Python 3.7.

If you find this project helpful, please give us a star ⭐️, your support is our greatest motivation.

Updates

🌟 2024.06.20 Revised the BiFA project.

Benchmark

Change detection methods supported by the repository

• FC-EF (ICIP'2018)
• FC-Siam-diff (ICIP'2018)
• FC-Siam-conc (ICIP'2018)
• IFN (ISPRS'2020)
• SNUNet (GRSL'2021)
• SwinUnet (TGRS'2022)
• BIT (TGRS'2022)
• ChangeFormer (IGARSS'22)
• MSCANet (JSTARS'2022)
• Paformer (GRSL'2022)
• DARNet (TGRS'2022)
• ACABFNet (JSTARS'2023)
• BiFA (TGRS'2024)
• ......

TODO

• Updated more change detection methods

Table of Contents

• Introduction
• Benchmark
• TODO
• Table of Contents
• Installation
• Dataset Preparation
• Model Training and Testing
• Citation
• License
• Contact Us

Installation

Requirements

• Linux system, Windows, depending on MMCV can be installed
• Python 3.6+, recommended 3.7
• PyTorch 1.10+ or higher, recommended 1.12.0
• CUDA 11.7 or higher, recommended 12.1

Environment Installation

It is recommended to use Miniconda for installation. The following commands will create a virtual environment named bifa and install PyTorch. In the following installation steps, the default installed CUDA version is 12.1. If your CUDA version is not 12.1, please modify it according to the actual situation.

Note: If you are experienced with PyTorch and have already installed it, you can skip to the next section. Otherwise, you can follow the steps below.

Step 0: Install Miniconda.

Step 1: Create a virtual environment named bifa and activate it.

conda create -n bifa python=3.7
conda activate bifa

Step 2: Install dependencies.

pip install -r requirements.txt

Note: Use the following command to install mmcv

pip install mmcv-full==1.6.2 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.12/index.html

Install BiFA

You can download or clone the BiFA repository.

git clone [email protected]:zmoka-zht/BiFA.git
cd BiFA

Dataset Preparation

Remote Sensing Change Detection Dataset

We provide the method of preparing the remote sensing change detection dataset used in the paper.

WHU-CD Dataset

• Data download link: WHU-CD Dataset PanBaiDu. Code:t2sb

LEVIR-CD Dataset

• Data download link: LEVIR-CD Dataset PanBaiDu. Code:qlvs

LEVIR+-CD Dataset

• Data download link: LEVIR+-CD Dataset PanBaiDu. Code: xtj8

SYSU-CD Dataset

• Data download link: SYSU-CD Dataset PanBaiDu. Code: gk0w

CLCD-CD Dataset

• Data download link: CLCD-CD Dataset PanBaiDu. Code: bcml

DSIFN-CD Dataset

• Data download link: DSIFN-CD Dataset PanBaiDu. Code: u8e9

Organization Method

You can also choose other sources to download the data, but you need to organize the dataset in the following format：

${DATASET_ROOT} # Dataset root directory, for example: /home/username/data/LEVIR-CD
├── A
│   ├── train_1_1.png
│   ├── train_1_2.png
│   ├──...
│   ├── val_1_1.png
│   ├── val_1_2.png
│   ├──...
│   ├── test_1_1.png
│   ├── test_1_2.png
│   └── ...
├── B
│   ├── train_1_1.png
│   ├── train_1_2.png
│   ├──...
│   ├── val_1_1.png
│   ├── val_1_2.png
│   ├──...
│   ├── test_1_1.png
│   ├── test_1_2.png
│   └── ...
├── label
│   ├── train_1_1.png
│   ├── train_1_2.png
│   ├──...
│   ├── val_1_1.png
│   ├── val_1_2.png
│   ├──...
│   ├── test_1_1.png
│   ├── test_1_2.png
│   └── ...
├── list
│   ├── train.txt
│   ├── val.txt
│   └── test.txt

Model Training and Testing

All configuration for model training and testing are stored in the local folder config

Example of Training on LEVIR-CD Dataset

python train.py --config/levir.json 

Example of Testing on LEVIR-CD Dataset

python test.py --config/levir_test_bifa.json 

BiFA Weight

• The model weights of BiFA are provided in the experiments/pretrain
• Segformer weight download link: Segformer PanBaiDu. Code:i81p

Citation

If you use the code or performance benchmarks of this project in your research, please refer to the following bibtex citation of BiFA.

@ARTICLE{10471555,
  author={Zhang, Haotian and Chen, Hao and Zhou, Chenyao and Chen, Keyan and Liu, Chenyang and Zou, Zhengxia and Shi, Zhenwei},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={BiFA: Remote Sensing Image Change Detection With Bitemporal Feature Alignment}, 
  year={2024},
  volume={62},
  number={},
  pages={1-17},
  keywords={Feature extraction;Task analysis;Remote sensing;Transformers;Interference;Decoding;Optical flow;Bitemporal interaction (BI);change detection (CD);feature alignment;flow field;high-resolution optical remote sensing image;implicit neural representation},
  doi={10.1109/TGRS.2024.3376673}}

License

This project is licensed under the Apache 2.0 License.

Contact Us

If you have any other questions❓, please contact us in time 👬