[1]杨 莉,崔丰韬,王婷婷.基于 YOLOv11n 的轻量化路面缺陷检测模型[J].机械与电子,2026,44(06):49-54.
　YANG Li,CUI Fengtao,WANG Tingting.A Lightweight Road Defect Detection Model Based on YOLOv11n[J].Machinery & Electronics,2026,44(06):49-54.
点击复制

基于 YOLOv11n 的轻量化路面缺陷检测模型()
分享到：

《机械与电子》[ISSN:1001-2257/CN:52-1052/TH]

卷:
44
期数:
2026年06期
页码:
49-54
栏目:
智能检测
出版日期:
2026-06-27

文章信息/Info

Title:
A Lightweight Road Defect Detection Model Based on YOLOv11n
文章编号:
1001-2257 ( 2026 ) 06-0049-06
作者:
杨 莉; 崔丰韬; 王婷婷
东北石油大学电气工程信息学院,黑龙江 大庆 163318
Author(s):
YANG Li ;  CUI Fengtao ;  WANG Tingting
( School of Electrical Engineering and Information , Northeast Petroleum University , Daqing 163318 , China )
关键词:
多尺度卷积; 路面缺陷检测; 深度学习; 高效检测头
Keywords:
multi-scale convolutions ;  road surface defect detection ;  deep learning ;  efficient detection head
分类号:
TP391
文献标志码:
A
摘要:
针对计算效率与检测精度之间的平衡问题,提出一种轻量化路面缺陷检测模型。该模型设计了 C3K2 _ MSCB 模块,采用多尺度卷积分支的并行设计,在低计算开销的同时增强特征提取能力。同时,引入 BiFPN-GLSA 模块,将双向特征金字塔网络与全局 局部自注意力机制相结合,实现了跨层级特征的高效融合。此外,采用 Detect _ Efficien 高效检测头,其轻量化解耦设计能够在分类与定位任务上进行更精细的特征学习,并显著降低参数量和推理延迟。在 RDD2022 数据集上进行评估,与 YOLOv11n 相比,改进模型在 mAP50 上提升 0.6 百分点,同时参数量、模型大小和 GFLOPs 分别降低 30% 、 23% 和 14% ,有效平衡了检测精度与计算效率。
Abstract:
To address the trade off between computational efficiency and detection accuracy , a lightweight road defect detection model is proposed.A C3K2 _ MSCB module is designed , which employs a parallel architecture based on multi-scale convolutional branches , enhancing feature extraction capabilities with low computational overhead.Concurrently , a BiFPN-GLSA module is introduced that integrates a bidirectional feature pyramid network with a global local self attention mechanism , achieving efficient cross level feature fusion.Furthermore , a Detect Efficient high efficiency detection head is adopted , and its lightweight decoupled design enables more refined feature learning for classification and localization tasks while significantly reducing parameter count and inference latency.Evaluated on the RDD2022 dataset , the improved model achieves a 0.6 percentage point increase in mAP50 compared with YOLOv11n , while simultaneously reducing the number of parameters , model size , and GFLOPs by 30% , 23% , and 14% , respectively , effectively balancing detection accuracy and computational efficiency.

参考文献/References:

[ 1 ] 交通运输部发布《 2024 年交通运输行业发展统计公报》[ J ] . 交通企业管理, 2025 , 40 ( 4 ): 42.

[ 2 ] Liu Zhuang , Li Jianguo , Shen Zhiqiang , et al.Learning efficient convolutional networks through network slimming [ PP / OL ] .arXiv ( 2017-08-22 )[ 2026-01-02 ] .http : ∥arxiv.org/ abs / 1708.06519.
[ 3 ] Redmon J , Divvala S , Girshick R , et al.You only look once : Unified , real-time object detection [ C ] ∥2016 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ) .Las Vegas , NV , USA : IEEE , 2016 : 779-788.
[ 4 ] Leng Jiaxu , Liu Ying.An enhanced SSD with feature fusion and visual reasoning for object detection [ J ] . Neural Computing and Applications , 2019 , 31 ( 10 ): 6549-6558.
[ 5 ] Girshick R.Fast R-CNN [ C ] ∥2015 IEEE International Conference on Computer Vision ( ICCV ) .Santiago , Chile : IEEE , 2015 : 1440-1448.
[ 6 ] He Kaiming , Gkioxari G , Dollar P , et al.Mask R-CNN [ C ] ∥2017 IEEE International Conference on Computer Vision ( ICCV ) .Venice : IEEE , 2017 : 2980-2988.
[ 7 ] Sun Zhen , Zhu Lingxi , Qin Su , et al.Road surface defect detection algorithm based on YOLOv8 [ J ] .Electronics , 2024 , 13 ( 12 ): 2413.
[ 8 ] Ren Xinghua , Hu Shaolin , Hou Yandong , et al.A lightweight convolutional neural network for detecting road cracks [ J ] .Signal , Image and Video Processing , 2024 , 18 ( 10 ): 6729-6743.
[ 9 ] Huang Yucheng , Liu Yuchen , Liu Fang , et al.A lightweight feature attention fusion network for pavement crack segmentation [ J ] .Computer-Aided Civil and Infrastructure Engineering , 2024 , 39 ( 18 ): 2811-2825.
[ 10 ] Liu Chang , Sun Yu , Chen Jin , et al.Improved lightweight road damage detection based on YOLOv5 [ J ] . Optoelectronics Letters , 2025 , 21 ( 5 ): 314-320.
[ 11 ] Wu Chenguang , Ye Min , Zhang Jiale , et al.YOLO-LWNet : a lightweight road damage object detection network for mobile terminal devices [ J ] .Sensors , 2023 , 23 ( 6 ): 3268.
[ 12 ] Tang Hanqi , Zhou Dandan , Zhai Haozhou , et al.RPD YOLO : a pavement defect dataset and real-time detection model [ J ] .IEEE Access , 2024 , 12 : 159738-159747.
[ 13 ] Wan Fang , Sun Chen , He Hongyang , et al.YOLO LRDD : a lightweight method for road damage detection based on improved YOLOv5s [ J ] .EURASIP Journal on Advances in Signal Processing , 2022 , 2022 ( 1 ): 98.
[ 14 ] Guo Jinbo , Wang Shenghuai , Chen Xiaohui , et al.QL YOLOv8s : precisely optimized lightweight YOLOv8 pavement disease detection model [ J ] .IEEE Access , 2024 , 12 : 128392-128403.
[ 15 ] Khanam R , Hussain M.YOLOv11 : an overview of the key architectural enhancements [ PP / OL ] .arXiv ( 2024-10-23 )[ 2025 12 03 ] .http : ∥arxiv.org / abs / 2410.17725.
[ 16 ] Rahman M M , Munir M , Marculescu R.EMCAD : efficient multi-scale convolutional attention decoding for medical image segmentation [ PP / OL ] .arXiv( 2024-05-11 )[ 2025-12-21 ] .http : ∥arxiv.org / abs / 2405.06880.
[ 17 ] Tan Mingxing , Pang Ruoming , Le Q V.EfficientDet : scalable and efficient object detection [ C ] ∥Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition ( CVPR ), 2020 : 10781-10790.
[ 18 ] Tang Feilong , Huang Qiming , Wang Jinfeng , et al.DuAT : dual aggregation transformer network for medical image segmentation [ PP / OL ] .arXiv ( 2022-12-21 )[ 2025-12-19 ] .http : ∥arxiv.org / abs / 2212.11677.
[ 19 ] Chen Jierun , Kao S H , He Hao , et al.Run , don ’ t walk : chasing higher FLOPS for faster neural networks [ C ] ∥ Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition , 2023 : 12021-12031.
[ 20 ] Arya D , Maeda H , Ghosh S K , et al.RDD2022 : a multi national image dataset for automatic road damage detection [ J ] .Geoscience Data Journal , 2024 , 11 ( 4 ): 846-862.
[ 21 ] Muhammad M B , Yeasin M.Eigen-CAM : class activation map using principal components [ C ] ∥2020 International Joint Conference on Neural Networks ( IJCNN ) . Glasgow , United Kingdom : IEEE , 2020 : 1-7.

相似文献/References:

[1]何正军,石雪敏,杜 涛,等.基于深度残差网络的轻量化电力敏感数据处理模型设计[J].机械与电子,2026,44(04):114.
　HE Zhengjun,SHI Xuemin,DU Tao,et al.Design of Lightweight Processing Model for Power-sensitive Data Based on Deep Residual Network[J].Machinery & Electronics,2026,44(06):114.

备注/Memo

备注/Memo:
收稿日期: 2026-03-11
基金项目:国家自然科学基金资助项目( 52474036 );黑龙江省自然科学基金资助项目( LH2024E008 )
作者简介:杨 莉 ( 1979- ),女,黑 龙 江 安 达 人,博 士,副 教 授,硕 士 研 究 生 导 师,研 究 方 向 为 智 能 控 制、最 优 控 制 和 神 经 网 络;崔丰韬 ( 2001- ),男,吉林吉林人,硕士研究生,研究方向为目标检测和深度学习,通信作者, E-mail :-1445318481@qq.com ;王婷婷 ( 1982- ),女,黑龙江安达人,博士,教授,博士研究生导师,研究方向为人工智能和信号处理。
更新日期/Last Update: 2026-08-26