计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (12): 175-188.doi: 10.11896/jsjkx.241200214

• 计算机图形学&多媒体 • 上一篇    下一篇

面向嵌入式应用的路面裂缝检测方法

胡鹏, 夏晓华, 钟预全   

  1. 长安大学道路施工技术与装备教育部重点实验室 西安 710064
  • 收稿日期:2024-12-30 修回日期:2025-05-06 出版日期:2025-12-15 发布日期:2025-12-09
  • 通讯作者: 夏晓华(xhxia@chd.edu.cn)
  • 作者简介:(18821791605@163.com)
  • 基金资助:
    国家自然科学基金(61901056);秦创原引用高层次创新创业人才项目(QCYRCXM-2022-352);陕西省交通运输厅科研项目(23-10X,24-74K,23-80X);陕西省重点研发计划(2024GX-YBXM-197)

Road Crack Detection Method for Embedded Applications

HU Peng, XIA Xiaohua, ZHONG Yuquan   

  1. Key Laboratory of Road Construction Technology and Equipment of MOE, Chang’an University, Xi’an 710064, China
  • Received:2024-12-30 Revised:2025-05-06 Published:2025-12-15 Online:2025-12-09
  • About author:HU Peng,born in 1999,postgraduate,is a member of CCF(No.V4622G).His main research interest is the designpostgraduate and application of artificial intelligence algorithms.
    XIA Xiaohua,born in 1987,Ph.D,professor,doctoral supervisor.His main research interests include machine vision and opto mechatronics integration.
  • Supported by:
    This work was supported by the National Natural Science Foundation of China(61901056),Qin Chuangyuan Cites High-Level Innovation and Entrepreneurship Talent Project(QCYRCXM-2022-352),Shaanxi Provincial Department of Transportation Scientific Research Project(23-10X,24-74K,23-80X) and Shaanxi Provincial Key Research and Development Program(2024GX-YBXM-197).

PDF (PC)

10

摘要： 针对基于深度学习的路面裂缝检测模型在嵌入式平台部署应用中存在模型复杂、处理速度慢等问题,基于YOLO提出一种面向嵌入式应用的路面裂缝检测方法。首先,在主干网络中构建两级串联卷积模块,优化卷积通道和输入空间的特征感知,并使用考虑预测框与真实框间向量角度的平滑交并比SIoU作为网络的损失函数,提高预测框回归的准确率和速率。其次,提出联级通道逐卷积剪枝的方法,先后进行通道剪枝和权重剪枝,移除非必要通道并逐卷积去除冗余权重,在保证模型性能稳定的同时显著压缩模型。接着,将YOLOv5l模型对裂缝特征的泛化能力蒸馏到剪枝后模型中,进一步提高模型对裂缝的表征能力。最后,在TensorRT引擎下,通过层优化等方式提高模型推理速度。实验结果表明,提出的方法与原模型相比,平均精度均值提高2.7%,模型参数量、模型体积分别减小78.24%和76.13%,检测速率提高587.88%;模型经TensorRT部署在NVIDIA Jetson Nano嵌入式设备上进行测试,在检测精度不受影响的同时,检测速率提高140%,与YOLOv5-Lite等常用模型及YOLOv10,RT-DERT等最新模型相比,具有最高检测精度以及最显著的轻量化效果,适合在嵌入式端应用。

关键词: 路面裂缝检测, 嵌入式应用, 两级串联卷积模块, 损失函数, 联级通道逐卷积剪枝, 知识蒸馏

Abstract: Aiming at the problems of complex models and slow processing speed of pavement crack detection model based on deep learning in embedded platform deployment and application,a pavement crack detection method for embedded application is proposed based on YOLO.Firstly,the two-stage concatenated convolutional module is constructed in the backbone network to optimize the feature perception of the convolution channel and the input space,and the Smooth Intersection over Union(SIoU) considering the vector angle between the prediction bounding boxes and ground truth bounding boxes is used as the loss function of the network to improve the accuracy and speed of the prediction bounding boxes regression.Secondly,a method of the cascade channel pruning and convolution-by-convolution weight pruning is proposed,and channel pruning and weight pruning are performed successively,the unnecessary channels are removed and the redundant weights are removed by convolution,which significantly compresses the model while ensuring the stability of the model.Then,distill the generalization ability of the YOLOv5l model for crack features into the pruned model to further improve its ability to characterize cracks.Finally,under the TensorRT engine,the inference speed of the model is improved through layer optimization and other methods.The experimental results show that compared with the original model,the mean Average Precsion of the proposed method is increased by 2.7%,the model parameters and model volume are reduced by 78.24% and 76.13% respectively,and the detection rate is increased by 587.88 %.The model is deployed on NVIDIA Jetson Nano embedded devices through TensorRT for testing,the detection accuracy is unaffected,and the detection rate is increased by 140%,compared with the commonly used models such as YOLOv5-Lite and the la-test models such as YOLOv10 and RT-DERT,it has the highest detection accuracy and the most significant lightweight effect,which is suitable for embedded applications.

Key words: Pavement crack detection, Embedded applications, Two-stage concatenated convolutional module, Loss function, Cascade channel pruning and convolution-by-convolution weight pruning, Knowledge distillation

中图分类号: 

  • U445.71
[1]WANG R,QI T Y.Research on crack features based on machine vision detection[J].China Civil Engineering Journal,2016,49(7):123-128.
[2]ZHANG K,CHEN D F,GAO M X,et al.Edge recognitionmethod for structural cracks in urban railway tunnels under supervised learning[J].China Mechanical Engineering,2021,32(4):446-453.
[3]KANG D,BENIPAL S S,GOPAL D L,et al.Hybrid pixel-level concrete crack segmentation and quantification across complex backgrounds using deep learning[J].Automation in Construction,2020,118:103291.
[4]ZHU J Q,ZHONG J T,MA T,et al.Pavement distress detection using convolutional neural networks with images captured via UAV[J].Automation in Construction,2022,133:103991.
[5]LUO H,JIA C,LI J.Highway pavement disease detection algorithm based on improved YOLOv4[J].Laser & Optoelectronics Progress,2021,58(14):336-344.
[6]HU H,LI Z,HE Z,et al.Road surface crack detection methodbased on improved YOLOv5 and vehicle-mounted images[J].Measurement,2024,229:114443.
[7]SHUVO M M H,ISLAM S K,CHENG J,et al.Efficient acce-leration of deep learning inference on resource-constrained edge devices:a review[J].Proceedings of the IEEE,2022,111(1):123-156.
[8]CHEN J,LIU Y,HOU J.A lightweight deep learning network based on knowledge distillation for applications of efficient crack segmentation on embedded devices[J].Structural Health Monitoring,2023,22(5):3027-3046.
[9]WU C G,YE M,ZHANG J L,et al.YOLO-LWNet:A lightweight road damage object detection network for mobile terminal devices[J].Sensors,2023,23(6):3268.
[10]HAO J M,YANG J Y,HAN S M,et al.YOLOv4 highway pavement crack detection method with the introduction of Ghost module and ECA[J].Journal of Computer Applications,2023,43(4):1284-1290.
[11]GENG H T,LIU Z Y,JIANG J,et al.Embedded road crack detection algorithm based on improved YOLOv8[J].Journal of Computer Applications,2024,44(5):1613-1618.
[12]LI P C,SUN L S,XIE Z W,et al.Road crack image detection algorithm based on improved MobileNet-SSD[J].Laser Journal,2022,43(7):123-127.
[13]ZHENG Y H,LIN X C,WU F,et al.A lightweight tobaccostem recognition method based on improved YOLOv4[J].Journal of Hefei University of Technology(Natural Science),2023,46(9):1196-1202,1253.
[14]GUO F,SUN X D,ZHU Q B,et al.Low resolution ceramic substrate image defect detection based on knowledge dis-tillation[J].Optics and Precision Engineering,2023,31(20):3065-3076.
[15]HINTON G,VINYALS O,DEAN J.Distilling the knowledge in a neural network[J].arXiv:1503.02531,2015.
[16]ZHANG L,MA K.Improve object detection with feature-based knowledge distillation:towards accurate and efficient detectors[C]//International Conference on Learning Representations.Addis Ababa,Ethiopia:OpenReview,2020.
[17]CHU J H,SHI L D,JING P G,et al.Context aware knowledge distillation network for object detection[J].Journal of Zhejiang University(Engineering Science),2022,56(3):503-509.
[18]CHO J H,HARIHARAN B.On the efficacy of knowledge distillation[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.IEEE,2019:4794-4802.
[19]ARASH A.AerialCrackDetection[EB/OL].http://github.com/arasharchor/AerialCrackDetection_Keras.
[20]REZATOFIGHI H,TSOI N,GWAK J Y,et al.Generalized intersection over union:A metric and a loss for bounding box regression[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2019:658-666.
[21]ZHENG Z,WANG P,REN D,et al.Enhancing geometric factors in model learning and inference for object detection and instance segmentation[J].IEEETransactions on Cybernetics,2021,52(8):8574-8586.
[22]YANG Z,WANG X,LI J.EIoU:an improved vehicle detection algorithm based on vehiclenet neural network[C]//Journal of physics:Conference series.IOP Publishing,2021.
[23]GEVORGYAN Z.SIoU loss:More powerful learning for bounding box regression[J].arXiv:2205.12740,2022.
[24]TONG Z,CHEN Y,XU Z,et al.Wise-IoU:bounding box re-gression loss with dynamic focusing mechanism[J].arXiv:2301.10051,2023.
[25]MA S,XU Y.Mpdiou:a loss for efficient and accurate bounding box regression[J].arXiv:2307.07662,2023.
[26]LIU C,WANG K,LI Q,et al.Powerful-IoU:More straightforward and faster bounding box regression loss with a nonmonotonic focusing mechanism[J].Neural Networks,2024,170:276-284.
[27]GUO L,LI R,JIANG B.A road surface damage detection method using Yolov4 with pid optimizer[J].International Journal of Innovative Computing,Information and Control,2021,17(5):1763-1774.
[28]SHIHAB U L M A,APSAAN M,AHAMED M H F,et al.An adaptive YOLO model for detection of faulty insulators in power transmission network using unmanned aerial vehicle[C]//2023 Moratuwa Engineering Research Conference(MERCon).IEEE,2023:282-287.
[29]CHEN X,GONG Z.Yolov5-lite:lighter,faster and easier to deploy[EB/OL].https://github.com/superrichiesui/YOLOv5-Lite.
[30]GE Z,LIU S,WANG F,et al.YOLOX:exceeding YOLO series in 2021[J].arXiv:2107.08430,2021.
[31]DUAN J,BAI T,LV B,et al.Lightweight detection of fasteners with YOLOv8 combined with ShuffleNetV2[C]//International Conference on Electrical and Information Technologies for Rail Transportation.Springer,2023:480-489.
[32]HUANG M,CAI Z.Steel surface defect detection based on improved YOLOv8[C]//International Conference on Algorithms,High Performance Computing,and Artificial Intelligence(AHPCAI 2023).IEEE,2023:1356-1360.
[33]WANG A,CHEN H,LIU L,et al.YOLOv10:Real-Time End-to-End object detection[J].arXiv:2405.14458,2024.
[34]LI M,WU J,WANG X,et al.Aligndet:aligning pre-training and fine-tuning in object detection[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.IEEE,2023:6866-6876.
[35]DU Z,HU Z,ZHAO G,et al.Cross-Layer feature pyramidtransformer for small object detection in aerial images[J].ar-Xiv:2407.19696,2024.
[36]ZHAO Y,LV W,XU S,et al.Detrs beat yolos on real-time object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2024:16965-16974.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发