A Sport Athlete Object Tracking Based on Deep Sort and Yolo V4 in Case of Camera Movement

• Published in: 2020 IEEE 6th International Conference on Computer and Communications (ICCC) pp. 1312 - 1316
• Main Authors: Zhang, Yao, Chen, Zhiyong, Wei, Bohan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 11.12.2020
• Subjects: Cameras; Deep Learning; Deep Sort; Games; Object detection; Object recognition; Object tracking; REID; Sports; Task analysis; Yolo V4
• DOI: 10.1109/ICCC51575.2020.9345010

Object tracking task has always been a major problem in the CV field. It is different from object detection. Object detection only needs to identify the type of object, while tracking task needs to identify its unique identity when a specific object is detected, such as REID problem. In sports-related fields, object tracking technology also has huge applications. For example, in football matches, camera tracking of footballs and tracking of athletes require this technology. This paper takes NBA and World Cup related scenes as the identification object, and aims to establish a tracking system for all players in the game, complete the realtime tracking of each athlete, so as to obtain relevant track information. In addition, the system can also help teachers review the competition in related education links and find the shortcomings of each student. Unlike most of filtering algorithms in past, this paper used the more cutting-edge deep learning technology in recent years, the YoloV4 and Sort's advanced version Deep Sort.