基于改进YOLOv5的驾驶员手持手机检测算法研究  

Research on Driver Handheld Phone Detection Algorithm Based on Improved YOLOv5

作　　者：彩朔 宋长明[1] CAI Shuo;SONG Changming(College of Science,Zhongyuan University of Technology,Zhengzhou 451191,China)

机构地区：[1]中原工学院理学院,河南郑州451191

出　　处：《现代信息科技》2023年第12期66-69,共4页Modern Information Technology

摘　　要：针对驾驶员手持手机行为检测精度低问题,提出一种改进的驾驶员手持手机行为检测算法。首先,在YOLOv5骨干网络中引入改进的注意力机制模块,更好地获取上下文信息,提高小目标检测的精确度。其次,采用一种改进的特征融合方法,提取三个尺度的特征,并对特征进行融合,更好地提取局部信息。实验结果表明,与YOLOv5相比,该检测算法在自制数据集上的精确度达到71.9%,提高了2.1%,对小目标的检测效果显著。An improved driver handheld phone behavior detection algorithm is proposed to address the issue of low accuracy in driver handheld phone behavior detection.First,an improved attention mechanism module is introduced into YOLOv5 backbone network to better obtain context information and improve the accuracy of small target detection.Secondly,an improved feature fusion method is adopted to extract features at three scales and fuse them to better extract local information.The experimental results show that compared with YOLOv5,the detection algorithm achieves an accuracy of 71.9%on the self-made dataset,get an improvement of 2.1%,which has a significant detection effect on small targets.

关 键 词：目标检测 YOLOv5 残差模块 注意力机制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]