Optical Flow Method-based Real Time Counting Method of Pull-up

Abstract

Pull-up is an important physical fitness measurement in various scenarios, such as high school entrance examination, firefighter's fitness test. Counting the number of pull-ups manually are labor exhausted and unreliable, many equipments are designed to solve this problem in an automatic manner. However, the existing solutions have three drawbacks: (1) The equipment is complex in some solutions, making them hard to deploy; (2) Some solution employ depth camera, which is expensive and cannot adapt to the variance of illumination; (3) the existing methods always need manual assistance. To address the aforementioned drawbacks, a real time pull up counting solution based on optical flow is proposed in this paper. We propose a pull-up counting solution named optical flow method-based real time counting method of pull-up. The pull-up counting can get test scores by analyzing real time video frames using optimal flow method and uses frame difference method to judge the criterion of action. To improve the time efficiency, restricted area, down-sampling, and taking frame interval operations are adopted. The test results show that the pull-up count method has a good performance, and the average test accuracy is 95.33%. The method can be applied to different test environments, having strong applicability.