PEBE691445/sat-image-boundingbox-sft-full

NU-TONIC原始SFT完整数据集包含卫星图像和对齐的土地覆盖输出，以图像-文本行的形式打包，用于SFT格式的微调。JSONL用户提示在重要时命名模态（卫星图像与空中背景）。数据来源包括GeoGuessr风格的POIs、Sentinel-2多光谱光学COGs（蓝/绿/红或视觉预览，百分比拉伸到uint8）、Earth Engine的土地覆盖栅格（重新投影到与RGB堆栈相同的10米网格，然后平铺并与掩码进行最近邻下采样）、可选的地理空中静止图像（基于令牌的静态地图API）等。数据集布局包括图像、覆盖层、地图背景静止图像、训练/验证/测试分割JSONL文件以及元数据。

The NU-TONIC raw SFT Full dataset contains satellite imagery and aligned land-cover outputs packaged as image–text rows for fine-tuning in SFT format. JSONL user prompts name the modality (satellite imagery vs. overhead context) where it matters. Provenance includes GeoGuessr-style POIs, Sentinel-2 multispectral optical COGs (blue/green/red or visual preview, percentile-stretched to uint8), per-pixel land-cover raster from Earth Engine (reprojected to the same 10 m grid as the RGB stack, then tiled and nearest-neighbor downsampled with the mask), optional geographic overhead still per POI (token-based static map API), etc. Layout includes images, overlays, mapbox stills, train/validation/test split JSONL files, and metadata.