Document Type



This study estimated stride length and frequency of runners in a 100 m race video. One method for measuring stride length and frequency is using infrared sensors. However, this method is not applicable to real races since numerous markers with infrared-reflective material must be attached to the runner’s entire body. Therefore, we proposed a method using a race video. We generated a panoramic image of the 100 m track to estimate the distance of each frame from the start line. We detected the positions of the runner’s steps from the movement of the leg joints. We projected every step to the overview image of the 100 m track. In the experiment, we applied our method to the video of an IAAF World Championship Track and Field 100 m race and obtained data from Usain Bolt. As a result, we can automatically estimate stride length and frequency of real races.