摘要
目前,市面上也已经有了很多的健身软件,但这些健身软件都只是单向信息输出,难以达到健身教练在旁指导的运动效果。本文利用Python的OpenCV进行图像分析,确定人体位置并实现骨骼关键点定位,采用MediaPipe确定人体的坐标位置并利用人体局部状态信息进行动作角度计算,根据预设定的准确度阈值比较判断动作是否正确,并给予相应分数作出相应反馈。
At present, there are also a lot of fitness software on the market, but these fitness software are only one-way information output, and it is difficult to achieve the exercise effect of fitness trainer beside guidance. In this paper, we use Python’s OpenCV to analyze the image, determine the position of the human body and locate the key points of the skeleton, Mediapipe is used to determine the coordinate position of the human body and calculate the action angle by using the local state information of the human body. According to the preset accuracy threshold, it can compare and judge whether the action is correct, and give the corresponding score to make the corresponding feedback.
作者
杨燕婷
张新松
马云莺
YANG Yanting;ZHANG Xinsong;MA Yunying(Zhicheng College Fuzhou University,Fuzhou Fujian 350002,China)
出处
《信息与电脑》
2022年第12期182-184,共3页
Information & Computer
基金
福建省大学生创新创业训练计划项目-AI健身指导(项目编号:S202113470008)。
关键词
人体关键点检测
视频图像处理
计算机视觉
human body key point detection
video image processing
computer vision
