摘要
目前,有很多基于机器学习的坐姿监测系统对用户的不规范坐姿进行预警,以减少由于不良坐姿造成颈肩腰椎发病问题。本文主要设计一种坐姿监测信息采集系统采集用于机器学习训练的数据,通过以坐在椅子上的用户作为对象,采用STM32为主控的压力传感器获取用户施加在椅子上的压力值,摄像头实时获取用户的坐姿视频流,识别人体关键点,收集人体姿态关键点的数据集,两种数据同时在基于树莓派的Pyqt界面上采集与整合。在获取不同坐姿情况下的传感数据的同时,用户可以借助视频画面中识别出的身体各部位关键点的实时位置,及时地将坐姿调整为本次采集数据对应的坐姿,避免在采集坐姿数据时,采集的数据并非用户此时对应的坐姿数据,从而确保采集的数据更加准确有效。结果表明,本系统结合传感器和视频流两种方式相较于仅使用单个采集数据可以更加快速、精确地采集用户坐姿信息数据。
Currently,there are many sitting posture monitoring systems based on machine learning to warn users of irregular sitting posture in order to reduce the incidence of neck,shoulder and lumbar spine due to poor sitting postures.This paper mainly designs a sitting posture monitoring information collection system to collect data for machine learning training.By taking a users sitting on the chair as the object,the pressure sensor controlled by STM32 is used to obtain the pressure value imposed by users on the chair.The camera obtains the user’s sitting posture video stream in real time,identifies the key points of human body,and collects the skeleton data set of human body posture.At the same time,the data were collected and integrated on the Pyqt interface based on Raspberry Pie.At the same time of obtaining the sensor data under different sitting posture,the user can adjust the sitting posture to the sitting posture corresponding to the data collected in time with the help of the real-time position of the key points of each part of the body identified in the video screen,so as to avoid that when collecting the sitting posture data,the collected data is not the sitting posture data corresponding to the user at this time,so as to ensure that the collected data is more accurate and effective.The results show that the system can collect the user’s sitting posture information more quickly and accurately by combining the sensor and video stream than by using only a single collection data.
作者
郭立祺
林涛
陈峰
叶炳旭
陈淑莹
王新家
陈耀文
GUO Liqi;LIN Tao;CHEN Feng;YE Bingxu;CHEN Shuying;WANG Xinjia;CHEN Yaowen(College of Engineering,Shantou University,Shantou 515063,China;Shantou Instrument Transformer CO.,LTD,Shantou 515041,China;The Second Affiliated Hospital of Shantou University Medical College,Shantou 515041,China;Central Laboratory,Shantou University,Shantou 515063,China)
出处
《中国体视学与图像分析》
2020年第1期72-81,共10页
Chinese Journal of Stereology and Image Analysis
基金
广东高校省级重大科研项目(2016KZDXM013)
2020年李嘉诚基金会交叉研究项目(2020LKSFBME06)
关键词
坐姿识别
Pyqt
薄膜压力传感器
sitting posture recognition
Pyqt
membrane pressure sensor
作者简介
郭立祺(1995-),男(汉),广东紫金,研究生。研究方向:信息与通信工程;通信作者:陈耀文,教授。研究领域:生物医学工程。E-mail:ywchen@stu.edu.cn
