SARDet-100K, DOTA, DroneVehicle
收藏资源简介:
该研究构建了一个多模态遥感目标检测的基准数据集,融合了SARDet-100K、DOTA和DroneVehicle三个数据集,涵盖了SAR、光学和红外等多种传感器模态。数据集旨在解决多模态遥感图像中的目标检测问题,特别是针对水平或旋转边界框的检测任务。通过整合不同模态的数据,该数据集为多任务学习提供了丰富的训练样本,能够有效提升模型在复杂场景下的泛化能力。数据集的应用领域包括无人机、卫星遥感等低空经济与地球观测任务,旨在实现多传感器数据的统一处理与高效检测。
This study constructs a benchmark dataset for multimodal remote sensing object detection, which integrates three datasets: SARDet-100K, DOTA, and DroneVehicle, covering multiple sensor modalities including SAR, optical, and infrared. This dataset is designed to solve object detection problems in multimodal remote sensing images, particularly detection tasks with horizontal or rotated bounding boxes. By integrating data from different modalities, it provides abundant training samples for multi-task learning, effectively improving the generalization capability of models in complex scenarios. The application fields of this dataset include low-altitude economy and Earth observation tasks such as unmanned aerial vehicle (UAV) and satellite remote sensing, aiming to realize unified processing and efficient detection of multi-sensor data.




