DOTA
收藏资源简介:
DOTA是由武汉大学创建的大规模航空图像目标检测数据集,包含2806张约4000×4000像素的图像,涵盖了15种常见目标类别,总计188,282个实例。数据集通过众包方式收集,由航空图像解释专家进行详细标注,每个实例使用任意(8自由度)四边形进行标记。DOTA旨在推动地球视觉(地球观测与遥感)中的目标检测研究,特别适用于解决航空图像中目标的尺度、方向和形状变化巨大的问题。
DOTA is a large-scale aerial image object detection dataset developed by Wuhan University. It comprises 2806 images with a resolution of approximately 4000×4000 pixels, covering 15 common object categories and containing a total of 188,282 annotated instances. The dataset was collected through crowdsourcing and meticulously annotated by aerial image interpretation experts, with each instance labeled using an arbitrary (8-degree-of-freedom) quadrilateral. DOTA is designed to advance object detection research in geovision (earth observation and remote sensing), and is particularly well-suited for addressing the challenges posed by extreme variations in object scale, orientation and shape in aerial images.




