Common Objects Underwater (COU)
收藏资源简介:
COU数据集是由明尼苏达大学计算机科学与工程系及明尼苏达机器人学院创建的,包含约9757张水下常见人造物体的实例分割图像。这些图像是从不同水下环境中收集的,包括游泳池、湖泊和海洋。数据集中的物体分为24类,包括海洋垃圾、潜水工具和自主水下航行器等。该数据集旨在解决水下实例分割数据集的缺乏问题,特别是为了训练轻量级、实时检测能力的自主水下航行器(AUVs)检测器。
The COU dataset was created by the Department of Computer Science and Engineering, University of Minnesota, and the Minnesota Robotics Institute. It contains approximately 9,757 instance segmentation images of common underwater man-made objects. These images were collected from various underwater environments including swimming pools, lakes, and oceans. The objects in the dataset are categorized into 24 classes, such as marine debris, diving equipment, autonomous underwater vehicles (AUVs), and others. This dataset aims to address the shortage of underwater instance segmentation datasets, specifically for training lightweight, real-time detection-capable autonomous underwater vehicle (AUV) detectors.




