基于轻量化 AWD-YOLOv8s的小目标螺栓检测算法
CSTR:
作者:
作者单位:

(1.湖北工业大学机械工程学院,武汉 430068;2.现代制造质量工程湖北省重点实验室,武汉 430068)

作者简介:

程博(1999-),男,硕士研究生,研究方向为目标检测、图像处理;王正家(1970-),男,博士,讲师,研究方向为物联网与大数据、机器视觉。

通讯作者:

中图分类号:

TP391. 4

基金项目:


Lightweight AWD-YOLOv8s-Based Algorithm for Detecting Small Target Bolts
Author:
Affiliation:

(1. School of Mechanical Engineering,Hubei University of Technology,Wuhan 430068,CHN; 2. Hubei Key Laboratory of Modern Manufacturing Quality Engineering,Wuhan 430068,CHN)

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对地铁车辆车顶图像中螺栓占比小,算法检测精度低和计算量大的问题,提出一种基于轻量化 AWD-YOLOv8s的小目标螺栓检测算法。首先,设计了一种自适应权重下采样 AWD模块,通过赋予局部特征信息不同的权重值,从而增强算法对小目标特征信息的提取能力;其次,引入尺度内特征交互(AIFI)模块替换 SPPF模块,通过捕捉同一尺度特征间的依赖关系,实现算法的轻量化改进;最后,采用轻量化 Ghost-P2结构,增强算法对小目标螺栓特征信息的提取能力。实验结果表明,改进后的算法平均精度均值达到 98. 7%,比原算法提高 2. 2%,模型参数量减少 46. 7%,检测速度达到 115 FPS,在实现模型轻量化的同时显著提高检测精度。

    Abstract:

    In this study,a small-target-bolt detection algorithm based on lightweight AWD-YOLOv8s is proposed to overcome the limitations of low detection accuracy and extensive computation in the analysis of a small proportion of bolts in subway-vehicle-roof images. First,an adaptive weight downsampling module is designed to enhance the ability of the algorithm to extract small-target feature information by assigning different weight values to local feature information. Second, an intrascale feature interaction module is introduced to replace the SPPF module. A lightweight algorithm with enhanced performance is obtained, which captures the dependency between same scale features. Finally,the lightweight Ghost-P2 structure is introduced to enhance the ability of the algorithm to extract the feature information of small target bolts. The experimental results show that the mean Average Percision of the improved algorithm has reached 98. 7%,which is 2. 2% higher than that of the original algorithm. In addition,the number of model parameters is reduced by 46. 7%,and a detection speed of 115FPS is achieved. Notably,the detection accuracy can be further improved while maintaining the“lightweight”feature of the model.

    参考文献
    相似文献
    引证文献
引用本文

程博,王正家,陈安澜,谷峰.基于轻量化 AWD-YOLOv8s的小目标螺栓检测算法[J].半导体光电,2026,47(1):203-209. CHENG Bo, WANG Zhengjia, CHENG Anlan, GU Feng. Lightweight AWD-YOLOv8s-Based Algorithm for Detecting Small Target Bolts[J].,2026,47(1):203-209.

复制
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2025-12-04
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2026-03-20
  • 出版日期:
文章二维码