SARDet-100K
收藏资源简介:
SARDet-100K数据集是由南开大学计算机科学与技术学院的研究团队开发的,旨在解决SAR目标检测领域数据集有限和代码不可访问的问题。该数据集通过整合和标准化10个公开的SAR检测数据集,提供了约116,598张图像和245,653个目标实例,覆盖了飞机、船只、汽车、桥梁、坦克和港口等6个类别。SARDet-100K不仅是首个COCO级别的大规模多类别SAR目标检测数据集,而且通过其大规模和多样性,为SAR目标检测算法的研究和评估提供了强有力的支持。数据集的标准化处理确保了图像分辨率和标注格式的统一,便于与流行的开源检测代码框架兼容,极大地促进了SAR目标检测技术的发展和创新。
The SARDet-100K dataset was developed by the research team from the College of Computer Science and Technology, Nankai University, aiming to address the challenges of limited available datasets and inaccessible code in the field of SAR object detection. This dataset integrates and standardizes 10 publicly available SAR detection datasets, providing approximately 116,598 images and 245,653 object instances covering 6 categories including aircraft, ships, cars, bridges, tanks, and ports. SARDet-100K is not only the first COCO-level, large-scale, multi-category SAR object detection dataset, but also offers robust support for the research and evaluation of SAR object detection algorithms through its substantial scale and rich diversity. The standardized processing of the dataset ensures consistent image resolution and annotation format, enabling compatibility with popular open-source object detection code frameworks, which greatly promotes the development and innovation of SAR object detection technologies.




