dca's Introduction

Deep Covariance Alignment (DCA)

Code for TGRS 2022 paper, "Deep Covariance Alignment for Domain Adaptive Remote Sensing Image Segmentation", accepted.

Authors: Linshan Wu, Ming Lu, Leyuan Fang

This repository highly depends on the LoveDA repository implemented by Junjue Wang. We thank the authors for their great work and clean code. Appreciate it!

Getting Started

Requirements:

pytorch >= 1.7.0
python >=3.6
pandas >= 1.1.5

Install Ever + Segmentation Models PyTorch + audtorch

pip install ever-beta==0.2.3
pip install git+https://github.com/qubvel/segmentation_models.pytorch
pip install audtorch

Prepare LoveDA Dataset

ln -s </path/to/LoveDA> ./LoveDA

Evaluate DCA Model on the test set

1. Download the pre-trained weights

2. Move weight file to log directory

mkdir -vp ./log/
mv ./URBAN_0.4635.pth ./log/URBAN_0.4635.pth
mv ./RURAL_0.4517.pth ./log/RURAL_0.4517.pth
python My_test.py

3. Evaluate on the website

Submit your test results on LoveDA Unsupervised Domain Adaptation Challenge and you will get your Test score.

Or you can download our results

Train DCA Model

python DCA_train.py

The training logs

Citation ✏️ 📄

If you find this repo useful for your research, please consider citing the paper as follows:

@ARTICLE{9745130,
  @ARTICLE{9745130,
  author={Wu, Linshan and Lu, Ming and Fang, Leyuan},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={Deep Covariance Alignment for Domain Adaptive Remote Sensing Image Segmentation}, 
  year={2022},
  volume={60},
  number={},
  pages={1-11},
  doi={10.1109/TGRS.2022.3163278}}

For any questions, please contact Linshan Wu.

Recommend Projects