基于YOLOv8的安全帽佩戴检测研究

Research on safety helmet wearing detection based on YOLOv8

随着安全生产意识的增强,工地安全监管日益受到重视,检测作业人员是否佩戴安全帽成为保障工地安全的一项重要措施。然而,安全帽的检测也存在不小的挑战,如存在目标尺寸变化、复杂背景干扰等因素。为此,文章提出了一种基于YOLOv8的安全帽佩戴检测方法,通过引入膨胀卷积以及卷积注意力机制,提升网络的特征提取能力,结合定位损失函数、置信度损失函数来进行参数的更新。实验数据显示,该方法的精度比原始的YOLOv8有一定的提升,可以准确地检测员工是否佩戴安全帽。

With the increasing awareness of safety production,construction site safety supervision is increasingly valued,and testing whether workers wear safety helmets has become an important measure to ensure the safety of construction sites.However,there are also significant challenges in the detection of safety helmets,such as changes in target size and complex background interference.This article proposes a safety helmet wearing detection method based on YOLOv8,which improves the network’s feature extraction ability by introducing dilated convolution and convolutional attention mechanism,and updates parameters by combining localization loss function and confidence loss function.The experimental data shows that the accuracy of this method has been improved compared to the original YOLOv8,and it can accurately detect whether employees are wearing safety helmets.