基于YOLOv6的未佩戴安全帽小目标检测方法研究  

Research on Small Target Detection Method for Unworn Helmet Based on YOLOv6

作　　者：马学友 李锋[1] 于守健[1] MA Xue-you;LI Feng;YU Shou-jian(College of Computer Science and Technology Computer,Donghua university,Shanghai 201620,China)

机构地区：[1]东华大学计算机科学与技术学院,上海201620

出　　处：《电脑与信息技术》2025年第1期1-5,26,共6页Computer and Information Technology

摘　　要：在施工环境下,检查工人是否佩戴安全帽是确保工作场所安全的重要任务,但传统方法费时费力。为解决这一问题,以YOLOv6作为基础模型进行改进。首先,采用深度可分离卷积(Depthwise Separable Convolution,DSC)对标准卷积进行修改,加速模型的识别速度,使其更适用于实时应用。其次,引入高效通道注意力(Efficient Channel Attention,ECA)模块,通过增强模型在特定区域的关注度,提高对小目标的识别精度。实验证明,改进的YOLOv6方法相较原始的YOLOv6,在识别速度上提高了24 ms,在识别精度上提升了2.1%,对施工环境下未佩戴安全帽的检测具有重要的实际意义。In the construction environment,checking whether workers wear helmets is an important task to ensure workplace safety,but the traditional method is time-consuming and laborious.In order to solve this problem,YOLOv6 was used as the basic model for improvement.Firstly,the standard convolution was modified by adopting Depthwise Separable Convolution(DSC)to accelerate the recognition speed of the model,making it more suitable for real-time applications.Secondly,the Efficient Channel Attention(ECA)module was introduced to enhance the recognition accuracy of small targets by increasing the model’s focus on specific areas.Experiments show that compared with the original YOLOv6,the improved YOLOv6 method improves the recognition speed by 24 ms and the recognition accuracy by 2.1%.It has important practical significance for the detection of unworn helmets in the construction environment.

关 键 词：安全帽 小目标 YOLOv6 深度可分离卷积 高效通道注意力 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]