Danish Airs and Grounds (DAG) dataset

Danish Airs and Grounds (DAG)数据集由丹麦数据供应与效率局创建，包含超过50公里的城市、郊区和农村地区的街景和航空图像，总计超过44,000张图像。该数据集主要挑战在于查询图像与参考图像之间极端的视角差异，以及随之而来的光照和透视变化。数据集创建过程中，首先从航空图像生成密集的3D重建，然后匹配查询的街景图像到3D模型的街景渲染。DAG数据集广泛应用于自主驾驶、机器人技术和增强现实等领域，旨在解决视觉定位和地点识别问题。

Danish Airs and Grounds (DAG) dataset was developed by the Danish Agency for Data Supply and Efficiency. It encompasses street view and aerial imagery covering over 50 kilometers of urban, suburban and rural areas, with a total of more than 44,000 images. The core challenges of this dataset arise from the extreme viewpoint disparities between query images and reference images, alongside concomitant illumination and perspective variations. During the dataset creation workflow, dense 3D reconstructions were first generated from aerial images, followed by matching the queried street view images to the street view renderings of the 3D model. The DAG dataset is widely applied in fields such as autonomous driving, robotics and augmented reality, aiming to address visual localization and place recognition tasks.