1e12Leon/UEMM-Air

UEMM-Air是一个大规模的合成无人机视觉数据集，包含超过120k个序列的6种配对多模态流和4D任务多样性，适用于感知、导航和自主性研究。数据集基于虚幻引擎打造，提供逼真的空中场景、精确的物理效果、多样的环境变化和像素级别的标注。支持多种任务，包括检测、分割、检索和跨模态理解，有助于全面感知建模。其合成性质确保了数据集的可扩展性、可重复性和对罕见事件的覆盖。

UEMM-Air is a large-scale synthetic drone vision dataset with 6 paired multimodal streams (over 120k sequences) and 4D task versatility, suitable for perception, navigation, and autonomy research. Built on Unreal Engine, it provides photorealistic aerial scenarios, precise physics, diverse environmental variations, and pixel-perfect annotations. Supporting multiple tasks including detection, segmentation, retrieval, and cross-modality understanding, it aids in comprehensive perception modeling. Its synthetic nature ensures scalability, reproducibility, and coverage of rare events.