EnglishVersion  
     
 
小组动态

LoveDA: A remote sensing land-cover dataset for domain adaptive semantic segmentation

1.Description

  The LoveDA (Land-cover Domain Adaptive semantic segmentation) dataset is a benchmark dataset designed for high spatial resolution (HSR) remote sensing land-cover semantic segmentation and unsupervised domain adaptation (UDA) tasks. It contains 5,987 HSR images with 166,768 annotated objects collected from three different cities. Unlike existing land-cover datasets that primarily focus on learning semantic representations, LoveDA introduces both urban and rural domains, posing significant challenges for model generalization. These challenges stem from multi-scale objects, complex background samples, and inconsistent class distributions, making it a valuable resource for developing robust and transferable segmentation models.


Fig. 1. LoveDA dataset

2.Annotation format

  Semantic category labels: background-1, building-2, road-3, water-4, barren-5,forest-6, agriculture-7, playground-8. The no-data regions are assigned 0. The following statistics provide an overview of the image distribution across different semantic categories in the LoveDA dataset.
Fig. 2. Statistics

3.Download

  We hope that the release of the LoveDA dataset can promote the development of remote sensing, land-cover classification task. You can click the link below to download the data.

● - Zenodo: download

● - Baidu Drive Code(27vc): download


4.Evaluation Server

  Submit your test results on LoveDA Semantic Segmentation Challenge , LoveDA Unsupervised Domain Adaptation Challenge.You will get your Test scores smoothly.

5.Copyright

  The owners of the data and of the copyright on the data are RSIDEA, Wuhan University. Use of the Google Earth images must respect the "Google Earth" terms of use. All images and their associated annotations in LoveDA can be used for academic purposes only, but any commercial use is prohibited.

[1] Wang, Junjue, Zhuo Zheng, Ailong Ma, Xiaoyan Lu, and Yanfei Zhong. "LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentation." In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track.
[2] Wang, Junjue, Zhuo Zheng, Ailong Ma, Xiaoyan Lu, and Yanfei Zhong. "LoveDA: A remote sensing land-cover dataset for domain adaptive semantic segmentation." arXiv preprint arXiv:2110.08733 (2021).

6.Contact

  If you have any the problem or feedback in using LoveDA dataset, please contact:
  Mr. Junjue Wang: kingdrone@whu.edu.cn
  Prof. Ailong Ma: maailong007@whu.edu.cn
  Prof. Yanfei Zhong: zhongyanfei@whu.edu.cn

 
版权所有@RS-IDEA | 地址:武汉市珞喻路129号 | 单位:武汉大学测绘遥感信息工程国家重点实验室 | 办公室:星湖楼709 | zhongyanfei@whu.edu.cn