SoccerNet-GSR

该数据集名为SoccerNet-GSR，是一个专注于足球视频的游戏状态重建新型数据集。它包含了200段各30秒的视频序列，每段视频都标注了球场定位、摄像机校准以及运动员的位置信息。此外，数据集中还包含了超过937万的线条点用于球场定位和摄像机校准，以及超过236万的运动员位置标注，这些位置信息关联了运动员的角色、所属球队和球衣号码。该数据集的规模为200个视频序列，其任务是进行游戏状态重建。

The dataset, named SoccerNet-GSR, is a novel dataset dedicated to game state reconstruction for soccer videos. It consists of 200 30-second video sequences, each annotated with field positioning data, camera calibration parameters, and player position information. Furthermore, the dataset contains over 9.37 million line points for field positioning and camera calibration, as well as more than 2.36 million player position annotations, which are linked to the player's role, affiliated team, and jersey number. With a total scale of 200 video sequences, the core task supported by this dataset is game state reconstruction.