无人机目标检测数据集

配备摄像头的无人机（或通用无人机）已迅速投入广泛应用领域，涵盖了农业、航空摄影、快速交付和监视等多个领域。因此，对于从这些平台收集的视觉数据的自动理解需求日益增加，使得计算机视觉与无人机技术之间的联系变得越来越密切。我们很高兴为各类重要的计算机视觉任务提供一个庞大的基准，精心标注了地面真实情况，将其命名为VisDrone，将视觉与无人机技术融合在一起。

Camera-equipped drones (or general-purpose unmanned aerial vehicles, UAVs) have been rapidly deployed across a wide spectrum of application domains, including agriculture, aerial photography, rapid delivery, surveillance and more. Consequently, the demand for automated understanding of visual data collected from these platforms has been on the rise, fostering increasingly tight integration between computer vision and UAV technology. We are pleased to introduce VisDrone: a large-scale benchmark meticulously annotated with ground truth, which combines visual and UAV technologies and supports a diverse array of critical computer vision tasks.

搜集汇总

背景与挑战

背景概述

该数据集是用于无人机目标检测的VisDrone2019-DET-train数据集，采用CC-BY-NC-SA-4.0许可证，可通过ModelScope SDK或Git LFS下载，需登录获取。