Remote Sensing Object Segmentation Dataset

The Remote Sensing Object Segmentation Dataset is a key asset for the remote sensing field, combining images from the DOTA open dataset and additional internet sources. With resolutions ranging from 451 × 839 to 6573 × 3727 pixels for standard images and up to 25574 × 15342 pixels for uncut large images, this dataset includes diverse categories like playgrounds, vehicles, and sports courts, all annotated for instance and semantic segmentation.

遥感目标分割数据集是遥感领域的一项关键资产，该数据集融合了来自 DOTA 开放数据集及额外互联网资源的图像。该数据集包含标准图像的分辨率从 451 × 839 像素至 6573 × 3727 像素，以及未经裁剪的大图像高达 25574 × 15342 像素，涵盖多样的类别，如游乐场、车辆和运动场地，并对实例分割和语义分割进行了标注。