YOLO-Rip
收藏资源简介:
YOLO-Rip数据集是由德国维尔茨堡大学计算机视觉实验室和罗马尼亚布加勒斯特大学合作创建的,包含2466张静态图像和17个无人机视频,总计约24K帧。这些图像和视频均经过专家验证,并带有精确的多边形注释,旨在为rip current的实例分割任务提供详尽的数据支持。数据集涵盖了不同时间、不同视角下rip current的形态,为研究者提供了丰富的学习样本,有助于提升海滩安全相关的检测技术。
The YOLO-Rip dataset was co-created by the Computer Vision Lab of the University of Würzburg, Germany, and the University of Bucharest, Romania. It comprises 2466 static images and 17 drone videos, totaling approximately 24,000 frames. All these images and videos have been expert-validated and paired with precise polygonal annotations, which are designed to provide comprehensive data support for rip current instance segmentation tasks. The dataset covers the morphological variations of rip currents across different times and perspectives, offering researchers abundant learning samples and facilitating the advancement of detection technologies related to beach safety.

- 1Rip Current Segmentation: A Novel Benchmark and YOLOv8 Baseline Results德国维尔茨堡大学计算机视觉实验室 · 2025年



