Multiple player tracking in basketball court videos
- PDF / 2,167,896 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 68 Downloads / 201 Views
SPECIAL ISSUE PAPER
Multiple player tracking in basketball court videos Xubo Fu1 · Kun Zhang2 · Changgang Wang2 · Chao Fan2 Received: 30 November 2019 / Accepted: 27 March 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract To build a smart basketball court, one basic task is to track players with the aid of the basketball court monitoring. This task can be regarded as a special case of the multiple object tracking (MOT) problem. But different from it, the task puts request to the existing MOT methods with both good accuracy and operational efficiency (toward real time) in the new basketball scenario. To deal with this task, we make the following attempts: (1) Considering the differences between pedestrians and basketball players and the lack of corresponding dataset for basketball players tracking, we construct a new MOT dataset under the basketball court monitoring scene, to better evaluate MOT methods in our task and to help promote future research in the related task, (2) evaluating the performance of the several candidate MOT methods on the new dataset, and (3) proposing the issues to be further addressed in this specific scenario. Keywords Smart basketball court system · Monitoring scene · Real time · Multiple object tracking
1 Introduction Basketball is an attractive and widely spread sport, and it is a great pleasure to get a sense of participation and to share the fun of playing. For the basketball crowd, texts and pictures obviously cannot perfectly present these wonderful moments, so videos are the best carriers. In response to these needs, the smart basketball court will have great application prospects. It can not only conveniently obtain sport data from installed cameras, but also intelligently analyze player behaviors, as well as automatically generate exciting highlights and sport data statistics. To achieve these goals, one basic task is to estimate trajectories of multiple players in each frame of a video sequence, with good accuracy and * Kun Zhang [email protected] Xubo Fu [email protected] Changgang Wang [email protected] Chao Fan [email protected] 1
Department of Public Physical and Art Education, Zhejiang University, Hangzhou 310058, China
Institute of Automation, Chinese Academy of Sciences (CASIA), Beijing 100190, China
2
running efficiency. In fact, this task belongs to the multiple object tracking (MOT) problem [6, 17, 28], a widely known task in the field of computer vision. But different from the conventional MOT problem, we focus on how to apply the MOT methods into the new scenario of basketball court, which is seldom investigated. Although there have been many advanced MOT methods emerging due to the advantage of deep neural networks [7, 16, 19, 22], there are still many difficulties to address the task of the basketball players tracking studied in this paper. We list these difficulties as follows: 1. Existing MOT methods [4, 7, 9, 16, 42] benefit a lot from the recent advances in object detection. For instance, the track
Data Loading...
