基于深度学习的口罩佩戴识别方法研究
Research on Mask Wearing Recognition Method Based on Deep Learning
叶柠 1孙宇舸 1张闯1
作者信息
- 1. 东北大学 信息学院,辽宁 沈阳 110819
- 折叠
摘要
针对自动检测口罩佩戴的问题,提出了一种基于深度学习的口罩佩戴识别方法.使用YOLOv3 网络为框架构建深度学习模型,并设计和完善了训练数据集.经过对网络模型的训练,成功解决了普通场景中单目标口罩识别问题.同时针对该模型在多目标检测中出现的检测置信度较低和部分目标无法识别等问题,进行了训练加强.通过对测试集的数据进行测试和分析,该方法具有较高的平均准确率(mean average precision,mAP)和稳定性.
Abstract
A mask wearing recognition method based on deep learning is proposed to address the issue of automatic detection of mask wearing.A deep learning model was constructed using the YOLOv3 network as the framework,and the training dataset was designed and improved.After training the network model,the problem of single target mask recognition in ordinary scenes has been successfully solved.At the same time,training was conducted to enhance the model's low detection confidence and inability to recognize some targets in multi target detection.By testing and analyzing the data from the test set,this method has a high average accuracy(mAP)and stability.
关键词
口罩佩戴识别/深度学习/YOLOv3Key words
mask wearing recognition/deep learning/YOLOv3引用本文复制引用
出版年
2024
NETL
NSTL
万方数据