Research on an Improved YOLOv7 Algorithm for Road Surface Crack Detection
Guangliang Weng, Dongsheng Liu, Linya Liu, Qianghua Wan, Songying Wu
East China Jiaotong University
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
This paper proposes an improved YOLOv7 algorithm for road surface crack detection to address the issues of low accuracy and inefficiency in complex background scenarios. Firstly, road surface crack sample data images with varying degrees of damage are collected in different road sections. The data is then processed using bilateral filtering and histogram equalization to create a high-quality road surface defect database. Next, efficient pyramid spatial attention (EPSA) modules and efficient spatial channel dual attention (SA) modules are embedded in the Backbone module of the original YOLOv7 network. Furthermore, improvements are made to the feature fusion path of the E-ELAN module by introducing skip connections and 1×1 convolutional structures between ELAB modules, enhancing the detection efficiency and overall performance of the network model. The preprocessed data is then used to train the improved YOLOv7 model, and the attention decision mechanism of the model is analyzed using Grad-CAM technology. Finally, a road surface defect intelligent detection system is developed based on the PyQt5 framework. Experimental results show that the improved YOLOv7 model achieves the highest recognition accuracy, recall rate, and mAP0.5 value for road surface cracks at 98.95%, 97.08%, and 98.97%, respectively. Compared to the original YOLOv7 model, the improved model demonstrates improvements in various evaluation metrics, meeting the requirements of road surface crack detection.
逐年被引趋势
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
工程Infrastructure Maintenance and Monitoring
Advanced Neural Network Applications · Industrial Vision Systems and Defect Detection