Improved Surface Defect Detection Model for Aluminum Profile Based on CSE-YOLO

LIU Zilong; PANG Ming; DOU Jianming

doi:10.11907/rjdk.251095

您当前的位置：

首页 >

文章列表页 >

Improved Surface Defect Detection Model for Aluminum Profile Based on CSE-YOLO

Artificial Intelligence | 更新时间：2026-04-30

- Improved Surface Defect Detection Model for Aluminum Profile Based on CSE-YOLO
- Software Guide Vol. 25, Issue 4, Pages: 98-103(2026)
- 作者机构：
  
  兰州交通大学机电工程学院，甘肃兰州 730070
- 作者简介：
- 基金信息：
- DOI：10.11907/rjdk.251095
  CLC： TP391.41
- Received：11 March 2025，
  
  Published：15 April 2026
- 稿件说明：
移动端阅览
刘子龙,庞明,窦建明.基于CSE-YOLO的改进铝型材表面缺陷检测模型[J].软件导刊,2026,25(04):98-103.

LIU Zilong,PANG Ming,DOU Jianming.Improved Surface Defect Detection Model for Aluminum Profile Based on CSE-YOLO[J].Software Guide,2026,25(04):98-103.
刘子龙,庞明,窦建明.基于CSE-YOLO的改进铝型材表面缺陷检测模型[J].软件导刊,2026,25(04):98-103. DOI： 10.11907/rjdk.251095.

LIU Zilong,PANG Ming,DOU Jianming.Improved Surface Defect Detection Model for Aluminum Profile Based on CSE-YOLO[J].Software Guide,2026,25(04):98-103. DOI： 10.11907/rjdk.251095.

摘要

针对提升铝型材表面缺陷检测精度时会导致模型的复杂度、检测速度与精度难以平衡的问题，提出一种基于改进YOLOv5的检测模型CSE-YOLO。首先，结合CCFF与BiFPN建立新的网络结构CCFBiFPN，以加强多尺度特征融合，降低模型的复杂度；其次，在特征融合模块引入混合GSConv的SDI模块，提高模型的特征提取能力；最后，在模型主干网络中增加ECA注意力机制，提高模型的精确度与总体性能。实验结果表明，改进模型相较于传统YOLOv5s模型在mAP@0.5和mAP@0.5：0.9指标上分别提升了5%与1.3%。同时，模型的参数量和浮点运算次数分别降低了4.3%与4.4%，且每秒检测帧数从140提升到了143。该方法有效平衡了模型的性能与复杂度，具有较高的工程应用价值。

Abstract

To address the challenge of balancing model complexity， detection speed， and accuracy in aluminum profile surface defect detection， an enhanced YOLOv5-based model named CSE-YOLO was developed. First， a novel CCFBiFPN architecture was constructed by integrating CCFF with BiFPN， which strengthened multi-scale feature fusion while reducing model complexity. Second， an SDI module incorporating hybrid GSConv was implemented in the feature fusion stage to improve feature extraction capability. Finally， an ECA attention mechanism was embedded in the backbone network to enhance detection precision and overall performance. Experimental results demonstrated that compared with the baseline YOLOv5s model， the proposed approach achieved 5% and 1.3% improvements in mAP@0.5 and mAP@0.5：0.9 metrics， respectively. Concurrently， parameter count and FLOPs were reduced by 4.3% and 4.4%， with the frame rate increasing from 140 to 143 FPS. The optimized architecture effectively balanced performance and complexity， demonstrating substantial practical value for industrial quality inspection applications.

关键词

Keywords

references

TAO X ， HOU W ， XU D . Overview of surface defect detection methods based on deep learning ［J］. IEEE/CAA Journal of Automatica Sinica ， 2021 （ 47 ）： 1017 - 1034 . doi: 10.16383/j.aas.c190811 http://dx.doi.org/10.16383/j.aas.c190811

陶显，侯伟，徐德 . 基于深度学习的表面缺陷检测方法综述［J］. 自动化学报， 2021 （ 47 ）： 1017 - 1034 . doi: 10.16383/j.aas.c190811 http://dx.doi.org/10.16383/j.aas.c190811

WU J L ， JIN Y Z . Research on surface defect detection of aluminum profile based on improved Faster R-CNN ［J］. Computer Era ， 2023 （ 11 ）： 52 - 57 .

吴吉灵，金玉珍 . 基于改进Faster R-CNN的铝型表面缺陷检测研究［J］. 计算机时代， 2023 （ 11 ）： 52 - 57 .

DENG G W ， YOU H Q ， ZHU Z S . Surface defect detection of aluminum profiles based on KCC-YOLOv ［J］. Laser & Optoelectronics Progress ， 2024 ， 61 （ 4 ）： 231 - 239 . doi: 10.3788/LOP230950 http://dx.doi.org/10.3788/LOP230950

邓光伟，尤红权，朱志松 . 基于KCC-YOLOv5的型材表面缺陷检测［J］. 激光与光电子学进展， 2024 ， 61 （ 4 ）： 231 - 239 . doi: 10.3788/LOP230950 http://dx.doi.org/10.3788/LOP230950

SUN L S ， WEI J X ， ZHU D M ， et al . Surface defect detection algorithm for aluminum profiles based on AM-YOLOv3 model ［J］. Laser & Optoelectronics Progress ， 2021 ， 58 （ 24 ）： 360 - 370 . doi: 10.3788/LOP202158.2415007 http://dx.doi.org/10.3788/LOP202158.2415007

孙连山，魏婧雪，朱登明，等 . 基于AM-YOLOv3型的铝型材表面缺陷检测算法［J］. 激光与光电子学进展， 2021 ， 58 （ 24 ）： 360 - 370 . doi: 10.3788/LOP202158.2415007 http://dx.doi.org/10.3788/LOP202158.2415007

TAN M X ， PANG R M ， LE Q V . EfficientDet： scalable and efficient object detection ［C］// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2020 ： 10778 - 10787 . doi: 10.1109/cvpr42600.2020.01079 http://dx.doi.org/10.1109/cvpr42600.2020.01079

ZHAO Y ， LYU W Y ， XU S L ， et al . DETRs beat YOLOs on real-time object detection ［C］// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2024 ： 16965 - 16974 . doi: 10.1109/cvpr52733.2024.01605 http://dx.doi.org/10.1109/cvpr52733.2024.01605

PENG Y ， SONKA M ， CHEN D Z . U-Net v2： rethinking the skip connections of U-Net for medical image segmentation［DB/OL］ . https：//arxiv.org/pdf/2311.17791 https://arxiv.org/pdf/2311.17791 . doi: 10.1109/isbi60581.2025.10980742 http://dx.doi.org/10.1109/isbi60581.2025.10980742

LI H L ， LI J ， WEI H B ， et al . Slim-neck by GSConv： a lightweight-design for real-time detector architectures ［J］. Journal of Real-Time Image Processing ， 2024 ， 21 （ 3 ）： 62 - 74 . doi: 10.1007/s11554-024-01436-6 http://dx.doi.org/10.1007/s11554-024-01436-6

WOO S ， PARK J ， LEE J-Y ， et al . CBAM： convolutional block attention module ［C］// Proceedings of the European Conference on Computer Vision ， 2018 ： 3 - 19 . doi: 10.1007/978-3-030-01234-2_1 http://dx.doi.org/10.1007/978-3-030-01234-2_1

WANG Q L ， WU B G ， ZHU P F ， et al . ECA-Net： efficient channel attention for deep convolutional neural networks ［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2020 ： 11534 - 11542 . doi: 10.1109/cvpr42600.2020.01155 http://dx.doi.org/10.1109/cvpr42600.2020.01155

Alibaba Group Holding Limite . 2018 Guangdong industrial intelligent manufacturing big data innovation competition - intelligent algorithm competition ［EB/OL］. https：//tianchi.aliyun.com/competition/entrance/231682/information https://tianchi.aliyun.com/competition/entrance/231682/information .

阿里巴巴集团控股有限公司 . 广东工业智造数据创新大赛—智能算法赛［EB/OL］. https：//tianchi.aliyun.com/competition/entrance/231682/information https://tianchi.aliyun.com/competition/entrance/231682/information .

HU J ， SHEN L ， SUN G . Squeeze-and-excitation networks ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence ， 2017 ， 42 ： 2011 - 2023 . doi: 10.1109/cvpr.2018.00745 http://dx.doi.org/10.1109/cvpr.2018.00745

LIU Y C ， SHAO Z R ， HOFFMANN N . Global attention mechanism： retain information to enhance channel-spatial interactions［DB/OL］ . https：//arxiv.org/pdf/2112.05561 https://arxiv.org/pdf/2112.05561 .

WANG C Y ， BOCHKOVSKIY A ， LIAO H M . YOLOv7： trainable bag-of-freebies sets new state-of-the-art for real-time object detectors ［C］// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2023 ： 7464 - 7475 . doi: 10.1109/cvpr52729.2023.00721 http://dx.doi.org/10.1109/cvpr52729.2023.00721

WANG A ， CHEN H ， LIU L H ， et al . YOLOv10： real-time end-to-end object detection［DB/OL］ . https：//arxiv.org/pdf/2405.14458 https://arxiv.org/pdf/2405.14458 . doi: 10.52202/079017-3429 http://dx.doi.org/10.52202/079017-3429

SUN T Q ， LIU J ， SONG C ， et al . Small target defect detection method for aluminum profile surface based on improved YOLOv8n ［J］. Modern Manufacturing Engineering ， 2024 （ 12 ）： 120 - 129 .

孙铁强，刘俊，宋超，等 . 基于改进YOLOv8n的铝型材表面小目标缺陷检测方法［J］. 现代制造工程， 2024 （ 12 ）： 120 - 129 .

XI L F ， YAERMAIMAITI Y ， LIU Y J . Surface defect detection method for aluminum profiles based on improved YOLOv5 ［J］. Journal of Guangxi Normal University（Natural Science Edition）， 2024 ， 42 （ 1 ）： 111 - 119 .

席凌飞，伊力哈木·亚尔买买提，刘雅洁 . 基于改进YOLOv5的铝型材表面缺陷检测方法［J］. 广西师范大学学报（自然科学版）， 2024 ， 42 （ 1 ）： 111 - 119 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on Improved YOLOV5 Algorithm for Dense Pedestrian Detection

Surface Defect Detection Algorithm of Butyl Rubber Stoppers Based on Improved YOLOv8

Unmanned Aerial Vehicle Image Small Target Detection Algorithm Based on SODM-YOLOv8

Integrated CNN and Transformer Dual-Branch Network for Pain Expression Recognition

A Source Code Summary Model Based on Feature Fusion and Copy Enhancement

Related Author

ZHOU Longgang

WEI Benchang

WEI Hongao

ZHANG Lu

LIU Yang

WU Fan

DING Xiaobo

ZHU Zhipeng

Related Institution

School of Electrical and Information Engineering， Hubei University of Automotive Technology

College of Computer and Information Technology， China Three Gorges University

Hubei Province Engineering Technology Research Center for Construction Quality Testing Equipment， China Three Gorges University

College of Computer Science and Engineering， Guilin University of Technology

College of Physics and Electronic Information Engineering， Guilin University of Technology

⁰