Research on Human Action Recognition Systems Based on Volleyball Match Cases

This article principally introduces the realization of a human posture recognition system. The recognition of human posture has a wide range of applications, including human-computer interaction, film and television production, motion analysis, and game entertainment, amongst others. Subsequently, the OpenPose algorithm is employed to facilitate the recognition of human posture, with the system primarily assessing human body movements by quantifying the distance between key points of human bones in successive frames of the video. The designed system achieved 70.5% accuracy in human keypoint detection on the test set, with an average recognition accuracy of around 60% for squatting, turning, and bending actions.