An efficient framework for analyzing periodical activities in sports videos

Abstract

In this paper, we propose a new and efficient framework for analyzing periodical activities in sports videos. The proposed framework first locates the motion patterns (MP) in each frame. By analyzing these located motion patterns with the proposed MP-analysis algorithm, the periodical sports activities can be effectively detected and counted. Furthermore, the qualified and unqualified actions can also be efficiently differentiated. Experimental results on various sports videos including dumbbell and weightlifting demonstrate the effectiveness of our proposed framework.