【摘要】 不良坐姿会使人产生背痛等症状，而久坐会增加心血管疾病的患病风险，甚至危及生命。坐姿监测与识别能够提醒人们久坐后活动，帮助纠正不良坐姿。设计了一个基于柔性压力传感阵列的坐姿监测与识别系统，包括坐姿采集、压力数据可视化、坐姿分类及状态显示与提醒四部分。对比分析了传统机器学习识别一维数组和轻量级卷积神经网络识别图像数据的效果，实现对正坐、左倾、右倾、前倾、后倾、左二郎腿、右二郎腿7种坐姿的精准识别，准确率可达99.83%。根据坐姿识别结果提醒使用者纠正坐姿，并结合计时器记录提醒使用者久坐后活动。更多还原

【Abstract】 Poor sitting postures cause symptoms such as back pain,and sedentary behavior increases the risk of cardiovascular disease and even be life-threatening.Sitting posture monitoring and recognition can remind people to move after sitting for a long time and help correct poor sitting posture.This research designed a sitting posture monitoring system based on a flexible pressure sensor array,which includes four parts:sitting posture collection,pressure data visualization,sitting posture classification,and status display and reminder parts.The performance of traditional machine learning for data of 1-D arrays and lightweight convolutional neural networks for the data of images is compared.Seven kinds of sitting postures are recognized,including sitting upright,leaning left,leaning right,leaning forward,leaning backward,cross left leg,and cross right leg.The best average accuracy in recognizing seven sitting postures reaches 99.83%.The final system reminds users to correct their sitting posture based on the results of sitting posture recognition,and timely remind users to move after prolonged sitting.更多还原