LoveDA
收藏资源简介:
LoveDA数据集是由武汉大学信息工程测绘国家重点实验室开发的一个高分辨率遥感地表覆盖数据集,包含来自三个不同城市的5987张图像和166768个标注对象。该数据集覆盖城市和农村两个领域,面临多尺度对象、复杂背景样本和类分布不一致等挑战。LoveDA数据集适用于地表覆盖语义分割和无监督域适应(UDA)任务,旨在解决大规模地表覆盖映射中的模型泛化问题。
The LoveDA Dataset is a high-resolution remote sensing land cover dataset developed by the State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University. It contains 5987 images from three different cities and 166,768 annotated objects. The dataset covers two domains: urban and rural areas, and faces challenges such as multi-scale objects, complex background samples, and inconsistent class distributions. The LoveDA Dataset is suitable for land cover semantic segmentation and unsupervised domain adaptation (UDA) tasks, aiming to address the model generalization issue in large-scale land cover mapping.

- 1LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentation武汉大学,信息工程测绘国家重点实验室 · 2022年



