HockeyAI
收藏资源简介:
HockeyAI数据集是一个专门用于推进冰球比赛中计算机视觉研究的开源数据集。该数据集包含约2100张高分辨率帧和详细的YOLO格式注释,提供了丰富的资源来解决快速运动环境中的目标检测挑战。数据集从瑞典冰球联赛(SHL)的广播视频中提取帧,并手动注释了动态对象(如球员、冰球)和静态冰场元素(如球门、中心冰)。数据集包含七个类别的注释,包括中心冰、争球点、球门、守门员、球员、冰球和裁判。数据集的分辨率为1920×1080像素,适用于研究人员、开发者和从业者改进冰球或类似动态场景中的目标检测和跟踪任务。
The HockeyAI Dataset is an open-source dataset specifically developed to advance computer vision research in ice hockey matches. It contains approximately 2100 high-resolution frames and detailed annotations in YOLO format, offering rich resources to tackle object detection challenges in fast-paced dynamic environments. All frames are extracted from broadcast videos of the Swedish Hockey League (SHL), with manual annotations covering dynamic objects such as players and hockey pucks, as well as static rink elements including goals and center ice. The dataset provides annotations for seven distinct categories: center ice, face-off spots, goals, goaltenders, players, hockey pucks, and referees. With a resolution of 1920×1080 pixels, the dataset is suitable for researchers, developers, and practitioners to improve object detection and tracking tasks in ice hockey or similar dynamic scenarios.




