当前位置>主页 > 期刊在线 > 计算机技术 >

计算机技术22年4期

一个基于 PaddlePaddle 深度学习框架的 口罩佩戴图像识别模型
谢美英
(湖南信息职业技术学院 软件学院,湖南 长沙 410100)

摘  要:后疫情时代的到来,口罩佩戴已经成为学校、商场、食堂等各公共场所的常规防控手段。为提高防控效率和准确率,研究搭建一个基于百度飞浆深度学习平台的口罩佩戴图像识别模型。该文先收集口罩佩戴人像图片样本,采用卷积神经网络VGG 算法训练模型,实现判断该静态人像是否规范佩戴口罩的检测功能,并对判断结果的准确性进行了评估。实验结果表明,VGG 卷积神经网络对判断人像是否佩戴口罩,具有比较快速、准确的识别能力,具备广泛应用的价值。


关键词:PaddlePaddle;卷积神经网络;VGG;口罩佩戴



DOI:10.19850/j.cnki.2096-4706.2022.04.029


中图分类号:TP391.4                                    文献标识码:A                                        文章编号:2096-4706(2022)04-0111-04



A Mask Wearing Image Recognition Model Based on PaddlePaddle Deep Learning Framework

XIE Meiying

(School of Software, Hunan College of Information, Changsha 410100, China)

Abstract: With the advent of the post epidemic era, mask wearing has become a routine means of prevention and control in schools, shopping malls, canteens and other public places. In order to improve the efficiency and accuracy of prevention and control, a mask wearing image recognition model based on Baidu flying slurry deep learning platform is studied and built. Firstly, this paper collects the picture samples of the mask wearing portrait, uses the convolutional neural network VGG algorithm to train the model, realizes the detection function of judging whether the static portrait wears the mask normally, and evaluates the accuracy of the judgment results. The experimental results show that VGG convolutional neural network has a relatively fast and accurate recognition ability to judge whether the portrait wears a mask, and has the value of wide application.

Keywords: PaddlePaddle; convolutional neural network; VGG; mask wearing


参考文献:

[1] GÉRON A. Scikit-Learn 与 TensorFlow 机器学习实用指南 [M]. 南京:东南大学出版社,2017.

[2] 邱锡鹏.神经网络与深度学习 [M].北京:机械工业出版社,2020.

[3] 娄茹珍,徐丽,蒋正乾,等 . 基于卷积神经网络的眼疾识别算法 [J]. 无线电工程,2021,51(11):1202-1207.

[4] 李亚可,玉振明 . 级联的卷积神经网络人脸检测方法 [J].计算机工程与应用,2019,55(24):184-189.

[5] 魏明珠,郑荣,杨竞雄 . 基于深度学习的图像检索研究进展 [J]. 情报科学,2021,39(5):184-192.

[6] 郑耀东,李旭峰 . 基于 Paddlepaddle 的新冠疫情期间文本情感分析模型 [J]. 信息与电脑(理论版),2021,33(7):189-191.


作者简介:谢美英(1984—),女,汉族,湖南涟源人,讲师,硕士研究生,研究方向:软件技术、数据挖掘、人工智能。