SoccerNet-Tracking
收藏资源简介:
SoccerNet-Tracking是一个专为足球视频中多目标跟踪设计的大型数据集,由列日大学和KAUST共同创建。该数据集包含200个30秒的视频序列,涵盖了12场足球比赛的精彩瞬间,以及一个完整的45分钟半场视频,用于长期跟踪评估。数据集中的视频均经过密集标注,包括球员轨迹、球队和球衣号码等信息。SoccerNet-Tracking旨在通过提供丰富的标注数据,推动足球视频分析领域的技术发展,特别是在球员跟踪、球权分析和战术理解等方面。
SoccerNet-Tracking is a large-scale dataset specifically designed for multi-object tracking in soccer videos, co-created by the University of Liège and KAUST. The dataset comprises 200 30-second video sequences covering highlight moments from 12 soccer matches, as well as one full 45-minute half-match video for long-term tracking evaluation. All videos in the dataset are densely annotated with information including player trajectories, team affiliations, and jersey numbers. SoccerNet-Tracking aims to advance technological developments in the field of soccer video analysis by providing rich annotated data, particularly in aspects such as player tracking, possession analysis, and tactical understanding.

- 1SoccerNet-Tracking: Multiple Object Tracking Dataset and Benchmark in Soccer Videos列日大学 · 2022年



