摘要
为解决口罩佩戴检测精度低、部署情况差等问题,提出基于人体关键点检测技术结合领域自适应迁移学习的口罩佩戴检测算法。将人脸图像进行关键点检测后进行图像分割,并对分割后的图像进行预处理;处理后使用数据增强技术并建立基于迁移学习的口罩佩戴检测模型;最终将训练好的模型嵌入至可移动设备中,实现口罩佩戴检测算法的部署。仿真结果表明,未经过人体关键点检测并人脸分割的口罩检测模型准确率为92.84%,经过人体关键点检测并分割图像的模型准确率为98.86%。实际硬件部署使用结果显示,经过人体关键点检测并分割图像的口罩佩戴检测精度超过95%。提出的算法在口罩佩戴检测中检测精度高,硬件部署情况好。
To solve the problems of low detection accuracy and poor deployment of masks,this paper presents a mask wearing detection algorithm based on human key point detection technology combined with domain adaptive mi-gration learning.Face image was segmented after key point detection,and the segmented image was preprocessed.Af-ter processing,data enhancement technology was used and a mask wearing detection model based on migration learning was established.Finally,the trained model was,embedded in the removable device to implement the mask wearing detection algorithm deployment.The simulation results show that the accuracy of the mask detection model without human key point detection and face segmentation is 92.84%,and that of the model with human key point de-tection and image segmentation is 98.86%.The actual hardware deployment usage results show that the wearing de-tection accuracy of the mask after human key point detection and image segmentation exceeds 95%.The proposed al-gorithm has high detection accuracy and good hardware deployment in mask wearing detection.
作者
叶永雪
马鸿雁
YE Yong-xue;MA Hong-yan(School of Electrical and Information Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Intelligent Processing for Building Big Data,Beijing 100044,China;National Virtual Simulation Experimental Center for Smart City Education,Beijing 100044,China)
出处
《计算机仿真》
北大核心
2023年第6期234-239,共6页
Computer Simulation
关键词
人体关键点检测
数据增强
迁移学习
口罩佩戴检测
硬件部署
Human key point detection
Data Enhancement
Transfer learning
Mask wear detection
Hardware Deployment
