RBOD:用于自动河道屏障目标检测的带注释的卫星影像数据集

全世界已经建造了数百万个河道屏障，用于防洪、水力发电和农业灌溉。缺乏关于河流屏障位置和类型的全面记录，特别是堰等小型屏障，限制了我们评估其社会和环境影响的能力。将卫星图像与物体检测算法相结合，有望在全球范围内自动识别河道屏障。然而，实现这一目标需要高质量的图像数据集进行算法训练和测试。因此，本研究提出了一个名为河流屏障物体检测（RBOD）的大规模数据集，使其成为第一个专门用于河道屏障物体检测的公开数据集。
RBOD数据集包括4,872张高分辨率卫星图像和11,741个精心注释的定向边界框（OBB）。在这个数据集中，河流屏障可以分为五类：水坝、沟壑、船闸、水闸和堰。使用五种典型的目标检测算法（YOLOV8-OBB、定向R-CNN、旋转更快的R-CNN、R3Det和旋转视网网）验证了RBOD数据集的有效性，为未来的应用提供了性能基准。

Millions of river barriers have been constructed worldwide for flood control, hydropower generation, and agricultural irrigation. The lack of comprehensive records on the locations and types of river barriers, especially small-scale ones such as weirs, limits our ability to assess their social and environmental impacts. Combining satellite imagery with object detection algorithms holds promise for automatically identifying river barriers at a global scale. However, achieving this goal requires high-quality image datasets for algorithm training and testing. Therefore, this study proposes a large-scale dataset named River Barrier Object Detection (RBOD), making it the first public dataset specifically dedicated to river barrier object detection.

The RBOD dataset includes 4,872 high-resolution satellite images and 11,741 meticulously annotated oriented bounding boxes (OBB). Within this dataset, river barriers are categorized into five classes: dams, gullies, locks, sluices, and weirs. The effectiveness of the RBOD dataset is validated using five typical object detection algorithms: YOLOV8-OBB, Oriented R-CNN, Rotated Faster R-CNN, R3Det, and Rotated RetinaNet, providing performance benchmarks for future applications.

搜集汇总

背景与挑战

背景概述

RBOD数据集是全球首个公开专门用于河道屏障目标检测的大规模卫星影像数据集，旨在通过目标检测算法自动识别全球河道屏障，以评估其社会环境影响。该数据集包含4872张高分辨率卫星图像和11741个标注的定向边界框，覆盖水坝、沟渠、船闸、水闸和堰五类屏障，数据经过专业标注和专家验证，适用于算法训练和测试。