改进YOLOv5的路面裂缝识别方法
Road crack detection method based on YOLOv5
杨景维 1黎远松1
作者信息
- 1. 四川轻化工大学计算机科学与工程学院,宜宾 644000
- 折叠
摘要
针对于深度学习在道路路面裂缝目标检测上检测精度低、错检、漏检等问题,提出一种基于改进YOLOv5的路面裂缝检测方法.首先在YOLOv5的主干网络中加入S2-MLPv2注意力模块,提高模型的精准定位能力.其次在颈部网络部分,使用GSConv+Slim-neck组合结构减轻模型复杂度的同时提升精度.最后将激活函数换为FReLU,在ReLU和PReLU的基础上进行改进,解决了激活函数中的空间不敏感问题.实验结果表明,与传统YOLOv5相比,本模型的mAP50增加了3个百分点,检测效果较好,能达到准确检测的要求.
Abstract
Aiming at the problems of low detection accuracy,error detection and missing detection in road surface crack target detection by deep learning,a road surface crack detection method based on improved YOLOv5 is proposed.Firstly,the S2-MLPv2 attention module was added to the backbone network of YOLOv5 to improve the model's accurate positioning ability.Secondly,in the neck part,GSConv+Slim-neck combined structure is used to reduce the complexity of the model and improve the accuracy.Fi-nally,the activation function is changed to FReLU,and ReLU and PReLU are extended to 2D activation function,which solves the space insensitivity problem in activation function.The experimental results indicate that compared to the traditional YOLOv5,this model achieves a 3-percentage-point increase in mAP50,demonstrating better detection performance and meeting the requirements for accurate detection.
关键词
裂缝检测/深度学习/YOLOv5/注意力机制Key words
crack detection/deep learning/YOLOv5/attention mechanism引用本文复制引用
出版年
2024
NETL
NSTL
万方数据