基于改进YOLOv8的道路工程空洞识别模型
A ROAD ENGINEERING VOID DETECTION MODEL BASED ON IMPROVED YOLOV8
杨卫朋1
作者信息
- 1. 中铁十一局集团城市轨道工程有限公司,武汉 430074
- 折叠
摘要
探地雷达因其无损检测和高效性特点,在地下缺陷诊断领域得到广泛应用.然而,现有的地下病害分类标准尚不明确,导致在道路地下病害检测过程中,往往需要依赖操作人员的主观判断,这不仅影响检测效率,也难以满足实际工程应用的高精度需求.为解决这一问题,文中提出一种基于改进YOLO(You Only Look Once)算法v8版本的地下空洞识别模型,在YO-LOv8模型中嵌入注意力机制,提高模型的检测速度以及准确度.通过使用杭州某路段的地下空洞数据对此模型进行验证.研究结果表明,改进后的YOLOv8模型在地下空洞检测方面表现出显著的性能提升,其中帧率(FPS)从26提升至29,平均精度(AP50)提高了1.7%,为地下病害诊断的自动化和智能化提供了新的解决方案.
Abstract
Ground penetrating radar(GPR)has gained widespread application in the field of underground defect di-agnosis due to its non-destructive testing and efficiency.However,existing classification standards for underground diseases are not clearly defined,leading to a reliance on the subjective judgment of operators during underground de-fect detection.This not only affects detection efficiency but also fails to meet the high precision requirements of prac-tical engineering applications.To address this issue,a model for identifying underground voids based on an im-proved YOLO(You Only Look Once)algorithm version 8 is proposed.An attention mechanism is embedded in the YOLOv8 model to enhance both the detection speed and accuracy.The model is validated using underground void data from a road section in Hangzhou.The results demonstrate a significant performance improvement of the im-proved YOLOv8 model in detecting underground voids,with the frame rate(FPS)increasing from 26 to 29 and the average precision(AP50)rising by 1.7%.This provides a new solution for the automation and intelligence of under-ground defect diagnosis.
关键词
地下病害检测/探地雷达/地下空洞/注意力机制Key words
underground disease detection/ground penetrating radar/underground cavity/attention mechanism引用本文复制引用
出版年
2024
NETL
NSTL
万方数据