Automated Judgment for Chin-Up Exercises Based on Microsoft Kinect Image System

Chin-up physical exercises need to be judged regular or foul and counted, the work can be automated finished by computer based on image recognition accurately and objectively. There are two difficulties in this automated system: one is to judge whether chin has been above the bar, the other is to determine whether chin-up arms have moment unbent state during sagging. This paper proposed judging chin above bar by Microsoft Kinect depth image and determining arm unbending by skeletal tracking. Depth image can help image segmentation, and skeletal tracking can position key points of arms. Experimental results showed that the approach could automate judge chin-up rapidly, accurately and practically.