[PDF] GLH-Water: A Large-Scale Dataset for Global Surface Water Detection in Large-Size Very-High-Resolution Satellite Imagery-论文阅读讨论-ReadPaper

摘要

Global surface water detection in very-high-resolution (VHR) satelliteimagery can directly serve major applications such as refined flood mapping andwater resource assessment. Although achievements have been made in detectingsurface water in small-size satellite images corresponding to local geographicscales, datasets and methods suitable for mapping and analyzing global surfacewater have yet to be explored. To encourage the development of this task andfacilitate the implementation of relevant applications, we propose theGLH-water dataset that consists of 250 satellite images and manually labeledsurface water annotations that are distributed globally and contain waterbodies exhibiting a wide variety of types (e.g., rivers, lakes, and ponds inforests, irrigated fields, bare areas, and urban areas). Each image is of thesize 12,800 $\times$ 12,800 pixels at 0.3 meter spatial resolution. To build abenchmark for GLH-water, we perform extensive experiments employingrepresentative surface water detection models, popular semantic segmentationmodels, and ultra-high resolution segmentation models. Furthermore, we alsodesign a strong baseline with the novel pyramid consistency loss (PCL) toinitially explore this challenge. Finally, we implement the cross-dataset andpilot area generalization experiments, and the superior performance illustratesthe strong generalization and practical application of GLH-water. The datasetis available at https://jack-bo1220.github.io/project/GLH-water.html.

展开全部

图表提取

论文十问

参考文献

被引用

提问与回答

暂无人提供速读十问回答