GSS²USW (Global-South Satellite&Street-view images for Urban Solid Waste classification) 全球南方典型城市固体废弃物分类卫星与街景影像数据集

We provide a spatially aligned multi-modal benchmark dataset specifically designed for identification of Urban Solid Waste (USW) in typical cities of the Global South, by integrating satellite remote sensing and panoramic street-view imagery, i.e., GSS2USW. Specifically, it covers Jakarta (Indonesia), Manila (Philippines), Dar es Salaam (Tanzania), Johannesburg (South Africa), Nakuru (Kenya), Medellín (Colombia), Belo Horizonte (Brazil), Santiago (Chile), and Buenos Aires (Argentina).Countries in the Global South commonly face governance challenges such as dense informal settlements, fragmented urban landscapes and severe illegal solid waste dumping. Nadir view from satellite remote sensing is highly susceptible to interference from tree occlusion and building shadows, while street-view imagery suffers from limited field-of-view and lacks macro-level scene context. To address these limitations, this dataset leverages the geographic coordinates of street-view sampling points as a linking anchor, and systematically constructs 5,145 spatially aligned multi-modal satellite and street-view samples, each consisting of a very-high-resolution satellite image patch and ground-level panoramic street view images. It aims to provide data support for low-cost, high-accuracy urban solid waste monitoring and multi-modal data fusion and computing. 本数据集是专门面向全球南方（Global South）典型城市固体废物（Urban Solid Waste, USW）点位级识别的卫星遥感与全景街景空间对齐多模态基准数据集。具体包括雅加达（印度尼西亚）、马尼拉（菲律宾）、达累斯萨拉姆（坦桑尼亚）、约翰内斯堡（南非）、纳库鲁（肯尼亚）、麦德林（哥伦比亚）、贝洛奥里藏特（巴西）、圣地亚哥（智利）和布宜诺斯艾利斯（阿根廷）。全球南方国家普遍面临非正规聚落密集、城市形态破碎、固废非法倾倒严重等治理难题。单一卫星遥感俯视视角极易受树木遮挡和建筑阴影干扰，而街景影像视距有限且缺乏宏观场景上下文信息。为此，本数据集以街景采样点地理坐标为纽带，系统构建了5145 组空间严格对齐的“高分辨率卫星影像切片 + 地面四视角透视街景”多模态样本对，旨在为低成本、高精度的城市固废监测与多模态遥感计算提供数据支撑。