基于多尺度特征的货运列车风管与折角塞门检测  

Detection of Duct Picking and Angle Cock of Train Based on Multi-scale Feature

作　　者：庞立波 孙林峰 唐飞 何晓晖 曾利平 石磊 Pang Libo

机构地区：[1]陕西榆横铁路有限责任公司,陕西榆林719000 [2]四川国软科技集团有限公司,四川成都610031

出　　处：《工业控制计算机》2024年第6期7-9,共3页Industrial Control Computer

基　　金：四川省经济和信息化厅-中国制造2025四川行动资金项目(CJXDZ-Y2021-003);陕西煤业化工集团有限责任公司2021年度板块级科研项目(2021SMHKJ-BK-J-52)。

摘　　要：针对小尺寸列车风管与折角塞门的检测问题,提出一种多尺度特征增强的YOLOv5s检测算法。首先在主干网络Backbone增加了一个大尺寸特征图输出分支,使用C2f(CSP bottleneck with 2 convolution and forward)模块丰富新增特征图的梯度流信息;然后在Neck部分使用自适应空间特征融合增强特征金字塔模型FPN (Feature Pyramid Network)的多尺度特征融合能力,并使用路径聚合模型PAN(Path Aggregation Network)提高目标定位能力;检测头Head拥有4种尺度的特征图,其中新增特征图的尺寸最大,用于增强小尺寸目标的检测能力。实验结果表明,多尺度特征增强的YOLOv5s比原来的YOLOv5s在平均精度mAP(mean Average Precision)上提升了1.5%,优于其他检测器,提升了小尺寸风管和折角塞门的检测能力。To improve the performance for detecting the duct pickings and angle cocks of train,an improved YOLOv5s based on multiple scale feature is proposed in this paper.Firstly,a new feature output is added at the backbone of YOLOv5s,where C2f block is exploited to enrich the gradient information and to detect the small object.The size of newly added feature map is doubled compared with that of original map.Then,an enhanced method for feature fusions is proposed at the Neck.Adaptively spatial feature fusion(ASFF)was utilized to improve the effectiveness of fused multiple scales of feature.Finally,the enhanced FAN with PAN framework was incorporated to fuse the feature maps of different size.Experimental results show that the proposed YOLOv5s achieves about 1.5%higher of mAP than the initial YOLOv5s,outperforming other compared methods.

关 键 词：目标检测 列车风管 折角塞门 多尺度特征 自适应空间特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U292.25[自动化与计算机技术—计算机科学与技术]