一种BiFPN-YOLOv8铁轨表面缺陷检测网络模型  

A BiFPN-YOLOv8 for Rail Surface Defect Detection Network Model

作　　者：冯庆贺 江铭凯 郝巧红 赵晓蕾 赵强 杨富超 FENG Qinghe;JIANG Mingkai;HAO Qiaohong;ZHAO Xiaolei;ZHAO Qiang;YANG Fuchao(School of Intelligent Engineering,Henan Institute of Techno logy,Xinxiang 453003,China;China Radiowave Propagation Research Institute,Qingdao 266107,China)

机构地区：[1]河南工学院智能工程学院,河南新乡453003 [2]中国电波传播研究所,山东青岛266107

出　　处：《河南工学院学报》2024年第3期25-28,43,共5页Journal of Henan Institute of Technology

基　　金：河南省重点研发与推广专项(科技攻关)项目(232102210094,222102210291);河南省国际科技合作项目(242102520036)。

摘　　要：针对铁轨表面存在的剥落、孔洞、瘢痕、划痕、裂缝、磨损缺陷,构建了一种BiFPN-YOLOv8铁轨表面缺陷检测网络模型。在数据准备中,通过Gauss噪声变换和水平变换进行铁轨表面图像数据增强;在网络构建中,通过在原YOLOv8网络模型中引入加权双向金字塔结构Bi FPN构建Bi FPN-YOLOv8网络模型用于提升铁轨表面缺陷检测性能;在实验仿真中,通过在RSDDs数据集上定量的对比和定性的检测结果评价,验证了BiFPN-YOLOv8网络模型在铁轨表面缺陷检测任务上的精准性和适用性。Aiming to defect such as peeling,holes,scars,scratches,cracks and wear on the rail surface,a BiFPN-YOLOv8 rail surface defect detection network model is constructed.In the data preparation phase,the rail surface image data is augmented through Gauss noise transformation and rotation transformation;In the network construction phase,in order to improve the performance of rail surface defect detection,the BiFPN-YOLOv8 is constructed by introducing the weighted bidirectional pyramid structure BiFPN into the original YOLOv8 network model;In the experimental simulation phase,the quantitative comparison and qualitative detection result evaluation on the RSDDs data set illustrates the accuracy and applicability of the BiFPN-YOLOv8 network model in the rail surface defect detection task.

关 键 词：铁轨表面 缺陷检测 BiFPN-YOLOv8 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]