基于改进YOLOv5的路面坑洞检测设计
Design of road pothole detection based on improved YOLOv5
周研逸 1周月娥 1沈琳芸 1沈立 1赵远东1
作者信息
- 1. 南京理工大学紫金学院,南京 210023
- 折叠
摘要
随着道路交通的增加,对于路面检测中目标的准确识别成为了一个重要的研究课题.提出了一种基于改进YOLOv5的目标检测算法,通过优化损失函数来提高路面检测中目标的识别精度.该设计算法对路面坑洞检测准确率达到了89.1%,相较于原始YOLOv5算法提升了7.7个百分点,同时出现漏检现象较少,具有较好的检测精度.结果表明,算法在目标检测任务中取得了较好的效果,准确性和实时性得到较高提升.
Abstract
With the increase of road traffic,the accurate identification of targets in pavement detection has become an impor-tant research topic.To enhance the recognition accuracy of targets in pavement detection,an improved YOLOv5-based target detec-tion algorithm is proposed in this design,which optimizes the loss function.In comparison to the original YOLOv5 algorithm,com-pared with the original YOLOv5 algorithm,the accuracy of road potholes detection reaches 89.1%surpassing the original YOLOv5 algorithm by a significant margin of 7.7%.At the same time,there were fewer missed detection phenomena,and the design algo-rithm had a good detection accuracy.The proposed algorithm yields excellent outcomes in the object recognition task,as demon-strated by the results,and its precision and real-time performance are enhanced.
关键词
目标检测/优化损失函数/路面检测/改进YOLOv5Key words
object detection/optimized loss function/road surface detection/improve YOLOv5引用本文复制引用
基金项目
江苏省大学生创新创业训练计划项目(202313654046T)
出版年
2024
NETL
NSTL
万方数据