Multimodal Detection of Dark Ships

This dataset contains a collection of videos consisting of satellite imagery augmented with 3D ship models, accompanied by the ships' corresponding AIS data. The intention of this dataset is for detecting dark ships, which are sea vessels acting maliciously, often while spoofing their AIS data. Multiple datasets exist that consist of satellite imagery of ships, however this dataset has the advantage of including each ships' corresponding AIS data. The simulated ships include both normal and anomalous behavior, whether the anomalous behavior is benign or malicious.Within the dataset are videos from a static view of PortMiami. The videos comprise 30 days of a 15-minute time window at the same time and of the same view of PortMiami. A single AIS data file is included, consisting of the AIS data for all ships visible in all 30 videos. There are also added columns in the AIS data to be used as ground truth information for different ship behaviors.The video data was generated using AGI's Simulation Tool Kit (STK).

本数据集汇集了一系列卫星图像，图像中融入了三维船舶模型，并附带相应的船舶自动识别系统（AIS）数据。该数据集旨在检测暗色船舶，即那些恶意行为的海上船只，通常在篡改其AIS数据时进行此类行为。尽管存在多个包含船舶卫星图像的数据集，但本数据集的独特之处在于它包含了每艘船舶对应的AIS数据。模拟船舶包括正常行为和异常行为，无论异常行为是良性的还是恶意的。数据集中还包含了来自迈阿密港静态视角的视频。这些视频包括在相同时间窗口内，对迈阿密港相同视角的30天内每15分钟的视频。此外，包含了一个AIS数据文件，其中包含所有30个视频中可见船只的AIS数据。在AIS数据中增加了额外的列，这些列将用作不同船舶行为的验证信息。视频数据是使用通用人工智能（AGI）的仿真工具包（STK）生成的。