公路病害自动检测及养护管理平台的设计与实现

Design and Implementation of Highway Disease Automatic Detection and Maintenance Management Platform

介绍了1种基于YOLO目标检测模型和元学习训练方法的软硬件系统平台,即道路养护管理平台,用于道路养护。针对我国的路面情况,道路养护管理平台设计了1套路面病害检测分类标准,用于构建道路状况数据库,并在真实路面进行了测试。基于YOLO目标检测模型采用低阶视觉和高阶视觉相结合的方法,将低阶视觉和高阶视觉数据进行融合,提高道路病害分析的鲁棒性。基于元学习的模型训练方法,保证模型可快速适应新的路面病害数据,提高模型泛化能力。实验结果表明,该平台设计的路面病害检测分类标准合理有效,所使用模型在路面病害分析方面具有较高准确性。本文提出的平台可为道路养护和管理提供有效的支持,并为在新技术背景下的国家道路病害标准提供依据。

This paper introduces a software and hardware system platform based on the YOLO target detection model and meta-learning training method,that is,a road maintenance management platform,which is used for road maintenance.According to the road condition in our country,the road maintenance management platform designed a set of road disease detection and classification standards used to build the road condition database and tested on the real road surface.Based on the YOLO target detection model,combining low-order vision and high-order vision is used to integrate low-order vision and high-order vision data to improve the robustness of road disease analysis.The meta-learning-based model training method ensures rapid adaptation to new pavement disease data and enhances the model s generalization capability.The experimental results demonstrate that the pavement disease detection and classification standard developed by the platform is both rational and effective.The model utilized exhibits high accuracy in pavement disease analysis,thereby offering valuable support for road maintenance and management.Furthermore,it serves as a foundation for national road disease standards amidst the backdrop of new technology.