Multi-complex-Seg

To develop effective and efficient multimodal image interpretation techniques, it is essential to comprehensively consider both phase and amplitude components of SAR data. Therefore, to facilitate advancements in multimodal segmentation in remote sensing, we propose and construct a multimodal terrain segmentation dataset, named Multi-Complex-Seg. Our dataset encompasses both urban and suburban areas and includes seven categories of land cover. These comprise six typical terrain types: water, road, grass, forest, farmland, and building, as well as an additional category labeled others. 为研发精准高效的多模态图像解译技术，需全面考量合成孔径雷达（Synthetic Aperture Radar，SAR）数据的相位与幅度分量。为此，为推动遥感领域多模态分割技术的发展，我们提出并构建了一款名为Multi-Complex-Seg的多模态地形分割数据集。该数据集覆盖城市与郊区区域，共包含七类土地覆盖类型，其中涵盖六种典型地形类别：水体、道路、草地、森林、农田与建筑，外加一类标注为“其他”的类别。