Visual tracking in camera-switching outdoor sport videos: Benchmark and baselines for skiing

• Published in: Computer vision and image understanding Vol. 243; p. 103978
• Main Authors: Dunnhofer, Matteo, Micheloni, Christian
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.06.2024
• ISSN: 1077-3142; 1090-235X
• DOI: 10.1016/j.cviu.2024.103978

Skiing is a globally popular winter sport discipline with a rich history of competitive events. This domain offers ample opportunities for the application of computer vision to enhance the understanding of athletes’ performances. However, this potential has remained relatively untapped in comparison to other sports, primarily due to the limited availability of dedicated research studies and datasets. The present paper takes a significant stride towards bridging these gaps. It conducts a comprehensive examination of skier appearance tracking in videos capturing their entire performance—an essential step for more advanced performance analyses. To implement this investigation, we introduce SkiTB, the largest and most annotated dataset tailored for computer vision applications in skiing. We subject a range of visual object tracking algorithms to rigorous testing, including both well-established methodologies and a novel skier-specific baseline algorithm. The results yield valuable insights into the suitability of various tracking techniques for vision-based skiing analysis and into the generalization of state-of-the-art algorithms to complex target behaviors and conditions set by winter outdoor environments. To foster further development, we make SkiTB, the associated code, and the obtained results accessible through https://machinelearning.uniud.it/datasets/skitb. [Display omitted] •Athlete localization is fundamental for high-level sport performance analysis.•Skiing is a popular winter sport discipline subject to extreme outdoor conditions.•We study visual tracking to localize a skier continuously in camera-switching videos.•We release SkiTB, the largest dataset for computer vision-based research in skiing.•We describe promising directions for further algorithmic development.