S²-Labeling: Shot-By-Shot Microscopic Badminton Singles Tactical Dataset

In this study, an affordable data collection solution is proposed for the tactical and strategic analysis of badminton. To make mass technical data collection possible, broadcast video is considered as the data source, and an intuitive and efficient video labeling tool is designed and implemented. The contribution of this work includes three folds. First, in response to the needs of the next generation of badminton technical and tactical analysis, the concept of Shot-By-Shot Labeling is proposed and a prototype data representation is formulated. Second, UI designs and computer vision techniques are utilized to develop an efficient computer-aided labeling tool for shot-by-shot technical data collection from match videos. Third, a Shot-By-Shot Labeling dataset composed of six badminton matches is shared with the community to accelerate related research. The dataset is available at https://drive.google.com/drive/folders/1eweBFa_M_CnYcqkQ5IHJ-Y8ag3TqRwUh.