SAR多域舰船目标检测数据集

数据主编：冷祥光; 谭向东; 计科峰; 匡纲要（国防科技大学电子科学学院，长沙 410073） 传统SAR舰船目标检测方法通常只适应于完全聚焦后的二维SAR图像。由于聚焦需要进行距离压缩、特别是距离徙动校正和方位压缩等一系列复杂且耗时的处理步骤，这使得目标检测的耗时始终受限于成像时间，很难实现SAR舰船目标的实时检测。针对上述问题，在SAR距离压缩域的舰船检测成为可行的方法。然而，目前仍然缺乏可靠的深度学习舰船检测数据集，能够满足距离压缩域上的检测。针对目前缺乏距离压缩域舰船目标样本数据的问题，本文介绍了RCship-1.0数据集。这是一个专门用于SAR距离压缩域舰船目标检测的数据集。该数据集是从公开的哨兵1号(Sentinel-1)卫星数据中选取若干SAR图像，采用逆线性调频变标（Inverse Chirp Scaling, ICS）算法得到距离压缩域数据。RCShip-1.0包含1000景大尺度SAR距离压缩域数据，为了便于网络训练，将大尺度图像直接切割成15210幅子图像，便于后续在大尺度SAR图像中呈现检测结果。在距离压缩域舰船目标检测数据集上的实验结果证明该数据集的可行性，标准性和公共可用性。 详细使用说明请参考RCShip-1.0数据集下载使用说明.pdf

Data Editors: Leng Xiangguang; Tan Xiangdong; Ji Kefeng; Kuang Gangyao (College of Electronic Science and Technology, National University of Defense Technology, Changsha 410073). Traditional SAR ship target detection methods usually only apply to fully focused 2D SAR images. Since the focusing process requires a series of complex and time-consuming operations including range compression, especially range migration correction and azimuth compression, the time cost of target detection is always constrained by the imaging time, making real-time detection of SAR ship targets difficult to implement. To address this issue, ship detection in the SAR range compressed domain has emerged as a feasible solution. However, there is currently a lack of reliable deep learning-based ship detection datasets that can support detection tasks in the range compressed domain. Aiming at the shortage of ship target sample data in the range compressed domain, this paper introduces the RCship-1.0 dataset, a dedicated dataset for SAR ship target detection in the range compressed domain. This dataset is generated by selecting multiple SAR images from public Sentinel-1 satellite data, and applying the Inverse Chirp Scaling (ICS) algorithm to obtain range compressed domain data. RCship-1.0 contains 1000 large-scale SAR range compressed domain images. To facilitate network training, these large-scale images are directly cropped into 15210 sub-images, which also facilitates the display of detection results in large-scale SAR images. Experimental results on the range compressed domain ship target detection dataset verify the feasibility, standardization and public availability of this dataset. For detailed usage instructions, please refer to the "RCship-1.0 Dataset Download and Usage Instructions.pdf".