摘 要:为了应对当下的疫情,加强公共安全防护,提高口罩佩戴检测精度,文章提出以目标检测 Anchor free 算法中的 Centernet 为基础,加入 CBAM 注意力机制的新网络模型,口罩检测的性能有所提升。对数据分为 face 即未戴口罩和 face_mask 即戴口罩两种类别,在公开数据集 VOC-MASK 上的实验表明,该目标检测的算法评估指标 MAP 值从 77.35% 到 79.87%增加了 2.52 个百分点,性能优于原始的 Centernet 算法。
关键词:口罩佩戴检测;Centernet;注意力机制
DOI:10.19850/j.cnki.2096-4706.2022.10.017
中图分类号:TP301.6 文献标识码:A 文章编号:2096-4706(2022)10-0067-06
Research on the Mask Wearing Detection Algorithm Based on Centernet
XING Chunyu1, ZHENG Ping2
(1.School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan 232001, China; 2.School of Electrical and Information Engineering, Anhui University of Science and Technology, Huainan 232001, China)
Abstract: In order to cope with the current epidemic situation, strengthen public safety protection and improve the accuracy of mask wearing detection. This paper proposes a new network model based on Centernet in object detection Anchor free algorithm and adding the CBAM attention mechanism. The performance of mask detection has been improved. The data are divided into two kinds of face (without masks) and face_mask (wearing masks). The experiments on public data set VOC-MASK show that the algorithm evaluation index Map value of the target detection increases by 2.52 percentage points from 77.35% to 79.87%, and its performance is better than the original Centernet algorithm.
Keywords: mask wearing detection; Centernet; attention mechanism
参考文献:
[1] DUAN K,BAI S,XIE L,et al. Centernet:Keypoint triplets for object detection [C]//Proceedings of the IEEE/ CVF International Conference on Computer Vision.Seoul: IEEE,2019:6569-6578.
[2] ZHOU X,WANG D,Krähenbühl P. Objects as points [J/OL].arXiv:1904.07850 [cs.CV].[2022-03-07].https://arxiv.org/abs/1904.07850.
[3]LAW H,DENG J. Cornernet:Detecting objects as paired keypoints [J].International Journal of Computer Vision,2020,128:642-656.
[4] 黄跃珍,王乃洲,梁添才,等 . 基于改进 CenterNet 的车辆识别方法 [J]. 华南理工大学学报(自然科学版),2021,49(7):94-102.
[5] 余旺旺 . 基于深度学习的快速人脸检测算法实现与应用研究 [D]. 成都:电子科技大学,2020.
[6] 薛均晓,程君进,张其斌,等 . 改进轻量级卷积神经网络的复杂场景口罩佩戴检测方法 [J]. 计算机辅助设计与图形学学报,2021,33(7):1045-1054.
[7] 叶冲,杨晶东 . 基于 CBAM-EfficientNet 的垃圾图像分类算法研究 [J]. 智能计算机与应用,2021,11(5):218-222.
[8] 郑学帅 . 基于 CenterNet 的交通标志检测与应用 [D]. 石家庄:河北师范大学,2021.
作者简介:邢春雨(1995—),女,汉族,安徽亳州人,硕士研究生在读,主要研究方向:图像处理。