YOLOv8模型在沥青路面隐性病害识别中的应用
Application of YOLOv8 Model in Recognition of Hidden Diseases in Asphalt Pavement
贾静静1
作者信息
- 1. 山西协力公路工程监理有限公司,山西 晋中 030600
- 折叠
摘要
探地雷达技术被广泛应用于路面隐性病害检测中,然而传统的雷达图像病害识别主要依赖人工,未形成标准化的流程,具有一定主观性且效率较低,不利于探地雷达技术的推广应用.采用YOLOv8深度学习模型对沥青路面病害进行学习和识别,相关试验结果表明,经过训练后的YOLOv8模型的平均精确率和召回率分别达到了 92.45％和89.25％,可以用于路面隐性病害的识别,并且将训练后的模型应用在某高速公路路面检测项目中,实现了对裂缝、层间黏结不良和松散3种病害的识别,研究方法及结果对于沥青路面隐性病害的检测具有较强的实际应用价值.
Abstract
Gound-penetrating radar technology has been widely applied in detecting hidden pavement diseases.However,conventional radar image recognition mainly relies on labor,without a standardized process,has a certain subjectivity and low efficiency,and is not conducive to promoting and applying ground-penetrating radar technology.This paper used the YOLOv8 deep learning model to study and recognize asphalt pavement diseases,and the study results showed that the average accuracy and recall rate of the trained YOLOv8 model reached 92.45％and 89.25％,respectively,and could be used in recognizing hidden pavement diseases.The trained model was applied in a detection project of the expressway pavement and three types of diseases were recognized,including cracks,poor cohesiveness between layers,and loose.The research method and results have strong practical application value for the detection of hidden diseases in asphalt pavements.
关键词
YOLOv8/探地雷达/病害检查/深度学习Key words
YOLOv8/ground-penetrating radar/disease detection/deep learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据