Soccerstar-v1-pc_uq.7z -

: Focuses on three primary event types: Goals , Yellow/Red Cards , and Substitutions .

: The paper proposes using recent developments in action recognition and detection to provide baselines, reaching a mean Average Precision (mAP) of 67.8% for classifying 1-minute temporal segments. SoccerStar-v1-pc_UQ.7z

The dataset was introduced by Silvio Giancola et al. at the CVPR 2018 Workshop on Computer Vision in Sports. It was designed to solve the problem of —temporally localizing sparse events like goals or cards within long video broadcasts. : Focuses on three primary event types: Goals

: It addressed the lack of large-scale, publicly available datasets for automated soccer video understanding, enabling the training of deep learning models for sports analytics. Evolutions of the Dataset at the CVPR 2018 Workshop on Computer Vision in Sports

Since the original v1 release, the dataset has expanded significantly into newer versions:

A Scalable Dataset for Action Spotting in Soccer Videos - arXiv