作者：李晓彤，程进，邹小平，李光东，王晓岚，任宝凯，王一飞，刘宝宇，王骏琦，周子肖，余兴
单位：北京信息科技大学北京市传感器重点实验室，北京 100101
中图分类号：TP391.41             
文献标识码：A             
文章编号：1006-883X(2021)01-0019-05
收稿日期：2020-10-28

摘要：近年来，我国儿童青少年的近视率呈上升趋势，引起近视的原因有很多，不正确的学习姿势也会增加近视发病的风险，可能还会影响儿童的成长发育。许多家长都十分重视儿童的学习坐姿问题，但却由于各种原因难以做到时刻的监督。针对这一问题，设计了一个基于人工智能的儿童坐姿监测系统。系统的核心为微型电脑，控制摄像头拍摄照片，再通过坐姿识别模型识别照片中的坐姿，在儿童保持低头等错误坐姿时，发出语音提醒其纠正。系统能利用卷积神经网络识别坐姿，及时且有针对性地提醒纠正，并且程序移植到了嵌入式平台，体积小巧，操作简便。
关键词：不良坐姿；人工智能；图像分类；语音提醒

A Monitoring System for Childrens Sitting Position Based on Artificial Intelligence
LI Xiaotong, CHENG Jin, ZOU Xiaoping, LI Guangdong, WANG Xiaolan, REN Baokai, WANG Yifei, LIU Baoyu, WANG Junqi, ZHOU Zixiao, YU Xing
Beijing Sensor Key Laboratory, Beijing Information Science & Technology University, Beijing 100101, China
Abstract: In recent years, the myopia rate of children and adolescents in China is on the rise. There are many reasons for myopia. Incorrect learning posture will increase the risk of myopia, and may also affect the growth and development of children. Many parents attach great importance to childrens learning posture, but for various reasons, it is difficult to achieve constant supervision for many of them. To solve this problem, this paper designs a monitoring system of childrens sitting position based on artificial intelligence. The core of the system is a micro computer, which controls the camera to take photos, and then recognizes the sitting position in the photos through the sitting posture recognition model. When the children maintain the wrong sitting position, such as lowering their head, they will be reminded to correct it by voice. The system can use the convolutional neural network to identify the sitting position, timely and targeted to remind correction, and the program has been transplanted to the embedded platform, small in size and easy to operate.
Key words: undesirable posture; artificial intelligence; image classification; voice remind

阅读全文

备注：2021年 第27卷 第01期