NightSongs / LightCDNet

LightCDNet: Lightweight Change Detection Network Based on VHR Images

Introduction

The Pytorch implementation for: "LightCDNet: Lightweight Change Detection Network Based on VHR Images" (IEEE GRSL' 2023)

You can get a PDF version of our paper here: Google Drive

Abstract

LEVIR-CD

Method	Crop Size	Lr schd	#Param (M)	MACs (G)	Precision	Recall	F1-Score	IoU
LightCDNet-small	256x256	40000	0.35	1.65	91.36	89.81	90.57	82.77
LightCDNet-base	256x256	40000	1.32	3.22	92.12	90.43	91.27	83.94
LightCDNet-large	256x256	40000	2.82	5.94	92.43	90.45	91.43	84.21

Notice

The code has been integrated into Open-CD, welcome to use it!

Environment installation

This project is implemented based on Open-CD, please refer to the installation method of Open-CD 0.x version.

Usage

(eg: LightCDNet-small)

Train:

python tools/train.py configs/lightcdnet/lightcdnet_s_256x256_40k_levircd.py --work-dir ./exp/lightcdnet_s_levir_workdir --gpu-id 0 --seed 602

Eval:

python tools/test.py configs/lightcdnet/lightcdnet_s_256x256_40k_levircd.py ./exp/lightcdnet_s_levir_workdir/latest.pth --eval mFscore mIoU

Visualization:

python tools/test.py configs/lightcdnet/lightcdnet_s_256x256_40k_levircd.py ./exp/lightcdnet_s_levir_workdir/latest.pth --format-only --eval-options "imgfile_prefix=tmp_infer"

Citation

If you find this project useful in your research, please consider cite:

@ARTICLE{10214556,
  author={Xing, Yuanjun and Jiang, Jiawei and Xiang, Jun and Yan, Enping and Song, Yabin and Mo, Dengkui},
  journal={IEEE Geoscience and Remote Sensing Letters}, 
  title={LightCDNet: Lightweight Change Detection Network Based on VHR Images}, 
  year={2023},
  volume={20},
  number={},
  pages={1-5},
  doi={10.1109/LGRS.2023.3304309}}