We present a method to analyze the sprinter's form in 100 m sprint races. To achieve good time records, sprinters should know the problem on their sprint form, for example, by analyzing the difference between good and bad forms. The proposed method captures two performance videos at different points (e.g., 50 and 90 m) in one 100 m sprint race, and matches them frame-by-frame to compare the two sprint forms. The method calculates the sprint motion similarity from body keypoints, and then matches video frames focusing on the periodicity of sprint motion and the linearity of sprint speed. We conducted experiments with performance video pairs for six sprinters, and confirmed that the proposed method could match them accurately, which showed the effectiveness and the usefulness of the method. Also, the coach of our track-and-field team could find some improvements of the sprinters by referring to the outputs of the method. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

中文翻译：

---

专注于运动周期性和速度线性的短跑表演视频匹配

我们提出了一种分析短跑运动员在 100 米短跑比赛中的状态的方法。为了获得良好的时间记录，短跑运动员应该知道他们的短跑形式上的问题，例如，通过分析好的和坏的形式之间的差异。所提出的方法在一场 100 米短跑比赛中在不同点（例如，50 米和 90 米）捕获两个性能视频，并逐帧匹配它们以比较两种短跑形式。该方法从身体关键点计算冲刺运动相似度，然后重点匹配冲刺运动的周期性和冲刺速度的线性度的视频帧。我们对六名短跑运动员的表演视频对进行了实验，并证实了所提出的方法可以准确地匹配它们，这表明了该方法的有效性和实用性。还，我们田径队的教练可以通过参考该方法的输出发现短跑运动员的一些改进。© 2021 日本电气工程师学会。由 Wiley Periodicals LLC 出版。