[1]许晓峰,陈姚节,刘 恒.基于 YOLO v3 的落水人员检测[J].计算机技术与发展,2022,32(08):49-54.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 008]
　XU Xiao-feng,CHEN Yao-jie,LIU Heng.Detection of People Falling into Water Based on YOLO v3[J].,2022,32(08):49-54.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 008]
点击复制

基于 YOLO v3 的落水人员检测()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年08期
页码:
49-54
栏目:
图形与图像
出版日期:
2022-08-10

文章信息/Info

Title:
Detection of People Falling into Water Based on YOLO v3
文章编号:
1673-629X(2022)08-0049-06
作者:
许晓峰1 ; 陈姚节1; 2 ; 刘 恒1
1. 武汉科技大学 计算机科学与技术学院,湖北 武汉 430065;
2. 智能信息处理与实时工业系统湖北省重点实验室,湖北 武汉 430065
Author(s):
XU Xiao-feng1 ; CHEN Yao-jie1; 2 ; LIU Heng1
1. Department of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan 430065,China;
2. Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System,Wuhan 430065,China
关键词:
YOLO v3; 聚类; 感受野模块; 注意力机制; 目标检测
Keywords:
YOLO v3; clustering; receptive field block; attention mechanism; target detection
分类号:
TP391
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 08. 008
摘要:
针对落水人员所处水域的复杂性以及波纹、阳光等因素导致对落水人员检测的准确率较低,以及在检测小目标时,经典检测算法易出现误检漏检情况,提出一种改进的 YOLO v3 目标检测算法。 使用 k-means++聚类算法对自有落水人员数据集进行聚类,得到更适合落水人员的锚框,从而提高检测速度与精度;在网络中加入通道注意力机制模块,其关注通道信息,可以学习到不同通道特征的重要程度,根据重要程度为每个通道分配相应的权重,从而让网络关注重要的特征,抑制不重要的特征,提高重要特征的表征能力;引入感受野模块( RFB) 来增大浅层特征图的感受野,从而提高小目标检测精度。 最后,在自制的落水人员数据集上对该算法进行了验证,结果表明,该算法在检测效果上优于原始 YOLO v3。
Abstract:
In view of the complexity of the water area where the person falling into the water and low accuracy of the detection of theperson falling into the water caused by the ripples,sunlight and? ?other factors,as well as false detection and missed detection of the classicdetection algorithm when detecting small targets,we propose an improved YOLO v3 target detection algorithm.? ? We use the k-means++clustering algorithm to perform clustering analysis on the data set of people falling into the water to obtain an anchor box that is moresuitable for people,thereby improving the detection speed and accuracy. A channel attention mechanism module,which is added in thenetwork and focuses on channel information,can learn the importance of different channel features and assign different weights for eachchannel based on the importance,so that the network pays attention to important features,suppresses unimportant features,and improvesthe characterization ability of important features. We introduce Receptive Field Module ( RFB) to increase the receptive field of theshallow feature map,thereby improving the accuracy of small target detection. Finally,the algorithm is verified on a self-made data set ofpeople falling into the water. It is showed that the proposed algorithm is superior to the original YOLO v3 in terms of detection effect.

相似文献/References:

[1]蒋璐璐 王适 王宝成 李慧敏 李鑫慧.一种改进的标记分水岭遥感图像分割方法[J].计算机技术与发展,2010,(01):36.
　JIANG Lu-lu,WANG Shi,WANG Bao-cheng,et al.Segmentation of Remote Sensing Image Based on an Improved Labeling Watershed Algorithm[J].,2010,(08):36.
[2]张甜 罗眉 孟晓红 赵宗涛.一种基于状态特征的航天发射故障诊断技术[J].计算机技术与发展,2010,(01):93.
　ZHANG Tian,LUO Mei,MENG Xiao-hong,et al.A Technology in Fault Diagnosis of Spaceflight Launch Based on State Character[J].,2010,(08):93.
[3]王会颖 章义刚.求解聚类问题的改进人工鱼群算法[J].计算机技术与发展,2010,(03):84.
　WANG Hui-ying,ZHANG Yi-gang.An Improved Artificial Fish- Swarm Algorithm of Solving Clustering Analysis Problem[J].,2010,(08):84.
[4]赵敏 倪志伟 刘斌.K—means与朴素贝叶斯在商务智能中的应用[J].计算机技术与发展,2010,(04):179.
　ZHAO Min,NI Zhi-wei,LIU Bin.Application Research of K - Means Clustering and Naive Bayesian Algorithm in Business Intelligence[J].,2010,(08):179.
[5]吴楠 胡学钢.基于聚类分区的序列模式挖掘算法研究[J].计算机技术与发展,2010,(06):109.
　WU Nan,HU Xue-gang.Research on Clustering Partition-Based Approach of Sequential Pattern Mining[J].,2010,(08):109.
[6]耿波 仲红 徐杰 闫娜娜.用关联分析法对负荷预测结果进行二次处理[J].计算机技术与发展,2008,(04):171.
　GENG Bo,ZHONG Hong,XU Jie,et al.Using Correlation Analysis to Treat Load Forecasting Results[J].,2008,(08):171.
[7]游芳 姜建国 张坤.基于二维属性的高维数据聚类算法研究[J].计算机技术与发展,2009,(05):111.
　YOU Fang,JIANG Jian-guo,ZHANG Kun.Cluster- Algorithm Studies Based on Two- Dimensional Attribute Higher - Dimension Data[J].,2009,(08):111.
[8]刘淑英 程国建 彭方.人工神经生长细胞结构网络在医疗诊断的应用[J].计算机技术与发展,2009,(05):231.
　LIU Shu-ying,CHENG Guo-jian,PENG Fang.Applications of Growing Cell Structures of Artificial Neural Network for Medical Diagnosis[J].,2009,(08):231.
[9]范新 沈闻 丁泉勋 沈洁.基于正例和未标文档的半监督分类研究[J].计算机技术与发展,2009,(06):58.
　FAN Xin,SHEN Wen,DING Quan-xun,et al.Research on Semi- Supervised Classification Based on Positive and Unlabeled Text Document[J].,2009,(08):58.
[10]王园园 倪志伟 赵裕啸 伍章俊.基于决策树的模糊聚类评价算法及其应用[J].计算机技术与发展,2009,(09):232.
　WANG Yuan-yuan,NI Zhi-wei,ZHAO Yu-xiao,et al.Fuzzy Clustering Evaluation Algorithm Based on Decision Tree and Application[J].,2009,(08):232.

更新日期/Last Update: 2022-08-10