基于YOLOv8的路面病害识别方法研究
Research on Pavement Distress Recognition Method Based on YOLOv8
王宏宇 1韩笑 1宋席发 1苏杰 1李杨 1吴雪静1
作者信息
- 1. 北华航天工业学院电子与控制工程学院,河北 廊坊 065000
- 折叠
摘要
路面病害自动识别是道路养护领域研究热点,其中如何提高路面病害自动识别率和精确度是研究者关注的重点.提出了基于YOLOv8 模型的路面病害识别方法,该方法利用YOLOv8 模型对图像预处理后的数据集进行训练,生成训练模型,以识别路面病害图像.实验结果表明,采用该方法进行路面病害识别,检测精度达到 94.6%,模型性能指标F1 值达到 0.94,取得了良好的识别效果.
Abstract
Automatic recognition of pavement damage is a research hotspot in the field of road maintenance,and how to improve the automatic recognition rate and accuracy of pavement damage is the focus of researchers.In this paper,the YOLOv8 model is used to identify pavement disease.This method uses the YOLOv8 model to train the data set after image pretreatment and generate a training model to identify pavement disease images.The experimental results show that the de-tection accuracy of this method is 94.6%the model performance index F1 value is 0.94,and the recognition effect is good.
关键词
路面检测/YOLOv8/VGG16/图像预处理Key words
pavement detection/YOLOv8/VGG16/image preprocessing引用本文复制引用
基金项目
河北省教育厅高等学校科研自筹经费项目(ZC2021004)
北华航天工业学院博士科研启动基金(BKY-2021-14/BKY2020-19)
北华航天工业学院科研项目(CXPT-2024-01)
出版年
2024
NETL
NSTL
万方数据