Badminton Rally Videos

该数据集包含了一系列羽毛球回合的视频片段，并对各种击球事件及其对应特征进行了详细标注。此外，数据集还引入了基于视频中单个击球预测的评分标准，其中包括对击球帧、击球者、球的高度等各项特征的详尽评估指标。该数据集的规模涵盖了多个回合视频及其相应的击球标注，任务旨在对羽毛球视频片段中的事件进行检测。

This dataset consists of a series of video clips of badminton rallies, with detailed annotations for various stroke events and their corresponding characteristics. Furthermore, it incorporates a scoring standard based on single-stroke prediction derived from the video, which includes exhaustive evaluation metrics for features such as the stroke frame, the player executing the stroke, and the height of the shuttlecock. The dataset encompasses multiple rally videos and their corresponding stroke annotations, and its core task is event detection within badminton video clips.

搜集汇总

背景与挑战

背景概述

该数据集是用于羽毛球击球事件检测的视频数据集，支持多种深度学习模型（如TrackNetV2、SwingNet、YOLO、ViT）进行事件识别和姿态估计。其特点在于专注于体育分析，结合了时序建模和计算机视觉技术，并公开了完整的训练和推理流程，适用于AI与体育结合的研究和应用。