基于深度学习的医院口罩佩戴检测系统研制
Development of Hospital Mask Wearing Detection System Based on Deep Learning
鄢小虎 1王姗 2翟金龙3
作者信息
- 1. 深圳职业技术大学 本科教育学院,广东 深圳 518055
- 2. 华中科技大学同济医学院附属武汉中心医院,湖北 武汉 430014
- 3. 辽宁科技大学 计算机与软件工程学院,辽宁 鞍山 114051
- 折叠
摘要
针对医院人流量大、检测时间短、干扰因素多的特点,对医院口罩佩戴检测系统的设计进行了研究.制作医院口罩佩戴检测数据集,利用LabelImg工具对口罩佩戴的位置和类别进行标注.为了减少光照对检测性能的影响,对图像进行白化处理.选择YOLOv5 模型进行检测,将迭代次数设置为不同的值,训练模型并比较评价指标值,找到合适的迭代次数.构建基于Flask和ECharts的展示页面,实现了对医院视频监控画面的实时检测,减少了误检和漏检,节省了人力物力.
Abstract
According to the characteristics of large flow of people,short detection time and many interfering factors,the design of hospital mask wearing detection system is studied.The detection data set of hospital mask wearing is made,and the position and category of mask wearing are marked by using the LabelImg tool.In order to reduce the influence of illumination on the detection performance,the image is whitened.The YOLOv5 model is selected for detection,the number of iterations is set to different values,and the model is trained and the evaluation index values are compared to find the appropriate number of iterations.The display page based on Flask and ECharts is constructed to realize real-time detection of hospital video surveillance pictures.It reduces false detection and missed detection,and saves manpower and material resources.
关键词
深度学习/口罩佩戴/目标检测/Flask框架Key words
Deep Learning/mask wearing/target detection/Flask frame引用本文复制引用
基金项目
国家自然科学基金(62102268)
深圳市高等院校稳定支持计划(20220812102547001)
深圳职业技术大学校级科研基金(6022312044K)
深圳职业技术大学校级科研基金(6024310045K)
深圳职业技术大学校级科研基金(6023310030K)
出版年
2024