Sidescan sonar images for training automated recognition of submerged body-like objects

This dataset contains sonar image data collected from various locations in Bath and Bristol, UK. It was collected using an autonomous uncrewed surface vessel (USV) equipped with Blueprint Subsea StarFish 450 and StarFish 990 side scanning sonar.

The higher resolution images were used to train convolutional neural networks (CNNs) for autonomous detection of a sunken mannequin, used as a proxy for a drowning victim in missing persons scenarios.

本数据集收录了自英国巴思和布里斯托尔等地收集的声纳图像数据。该数据集系采用配备Blueprint Subsea StarFish 450及StarFish 990侧扫声纳的自主无人水面舰艇（USV）采集而成。其中，高分辨率图像被用于训练卷积神经网络（CNNs），以实现对于沉没的模拟人偶的自主检测，该人偶作为溺水受害者的代理，在失踪人员场景中应用。