摘要
为了解决目前使用的坐姿识别技术存在成本高、操作复杂等问题,设计一种柔性传感单元并制成柔性传感阵列坐垫,提出了一种深度学习网络结构,通过实验对5种不同的坐姿进行识别,平均准确率达到98.46%。该方法将柔性传感坐垫与深度学习结合,提供了一种成本低、普适性强的方案,有助于提示青少年及久坐人群的坐姿问题,从而改善因坐姿问题导致的并发症。
in order to solve the problems of high cost and complex operation in the current sitting posture recognition technology,a flexible sensing unit is designed and made into a flexible sensing array cushion,and a deep learning network structure is pr oposed.Five different sitting postures are recognized through experiments,and the average accuracy is 98.46%.This method combines the flexible sensing cushion with deep learning to provide a low-cost and universal scheme,which is helpful to prompt the sitting posture problems of teenagers and sedentary people,so as to improve the complications caused by sitting posture problems.
作者
陈浩川
贾康昱
胡新荣
CHEN Hao-chuan;JIA Kang-yu;HU Xin-rong(School of Computer and Artificial Intelligence,Wuhan Textile University,Wuhan Hubei 430200,China;Key Laboratory of Textile Fiber and Products,Wuhan Textile University,Wuhan Hubei 430200,China)
出处
《武汉纺织大学学报》
2022年第3期3-8,共6页
Journal of Wuhan Textile University
关键词
柔性传感阵列
坐姿识别
深度学习
flexible sensor array
sitting posture recognition
deep learning
